To diet or not to diet – that is still the question

To identify differences between malignant and normal cells is of great importance for tumor therapies, which are intended to selectively inhibit the growth of tumors while sparing healthy tissues. In this context, nearly 100 years ago, one of the first observations in tumor research was that tumor metabolism differs from that of differentiated cells: transformed cells consume much more glucose and produce high amounts of lactate even in the presence of oxygen (aerobic glycolysis).¹ While in the following decades, genetics and molecular biology dominated the mainstream of tumor research, interest in tumor metabolism has been refreshed in recent years—partly because molecular targeted therapies have shown disappointing results in a large proportion of solid tumors–and the molecular basis for the alterations in tumor metabolism is becoming increasingly better understood.

The phenomenon that increased glucose uptake can be observed in nearly all tumor types highlights the fundamental role of aerobic glycolysis for tumor growth^2,3 and makes it therefore a rational target for novel therapies; also conversely, inhibitors of glycolysis or lactate transporters are currently being developed and tested in their first clinical trials⁴ (eg, NCT01791595). However, many obstacles regarding specificity and toxicity need to be overcome and so far no compound has entered Phase III trials.

An alternative approach to target increased glucose needs of tumors is the restriction of glucose availability. Such a dietary approach has appealing theoretical advantages over drug-based therapies: solid tumors are characterized by impaired vascularization and hypoxia. While the low vascularization impairs delivery of drugs into the tumor such that only cells near blood vessels come in contact with the substance, lowering blood levels of nutrients (such as glucose and amino acids) might be especially effective in poorly vascularized areas, adding a novel level of tumor selectivity and making such restrictive approaches ideal candidates for combinatorial therapies.

One way to restrict glucose by dietary means is the use of low carbohydrate ketogenic diets (KDs). A number of important studies explored the efficacy of KDs in different mouse models.^5–7 In most of these studies, a KD did impair tumor growth in mouse xenograft experiments, and efficacy often was related to the decrease in plasma glucose.^5,7 Lowering blood glucose levels leads to the generation of ketone bodies such as beta-hydroxybutyrate (BHB), which serve as alternative nutrients to different organs such as the brain and heart.⁸ Although a number of in vitro studies showed that established tumor cell lines did not metabolize ketone bodies, and immunohistochemical analyses demonstrated that a large proportion of gliomas do not express enzymes necessary for the metabolism of ketones,⁹ bona fide analyses of ketone metabolism in glioma in vivo were lacking.

In this issue of Neuro-Oncology, De Feyter and colleagues now present such data. By using 2 different rat glioma cell lines, they confirmed the finding that these cells do not substantially metabolize BHB in vitro. Surprisingly, using infusion of ¹³C-labeled BHB and sophisticated MRI spectroscopy, they carefully analyzed BHB metabolism in rat xenograft models and found that glioma tissue in vivo took up BHB and metabolized it to glutamate in a way similar to that of healthy cortex. Further, the amount of BHB accumulation in tumors was higher in rats on a KD, suggesting that these tumors upregulated BHB transporters when they were grown in a ketotic environment. Indeed, the authors found upregulation of the monocarboxylic acid transporter 1, which transports ketones and lactate among other substances. Finally, the KD did not slow the growth of orthotopic 9L or RG-2 tumors.

This is the first study to investigate BHB metabolism in vivo during a KD in glioma, and its results challenge current concepts using KD as tumor therapy. However, while the study elegantly demonstrates the metabolic flexibility of glioma cells, it also invites new questions. First, the finding that the KD did not inhibit tumor growth although it strongly reduced plasma glucose levels is in contrast to the majority of mouse xenograft models. Can this be attributed to species differences, as it is known that ketone metabolism differs between mouse, rat, and man?^10–13

Second, why was ketone metabolism detected in vivo but not in vitro? Are nontumoral cells responsible for the BHB uptake in the xenograft tumors? Do microenvironmental adaptations in vivo induce BHB metabolism? If gliomas during a KD truly undergo an evolution toward ketone metabolism and oxidative phosphorylation, does this confer susceptibility toward mitochondrial stress and hypoxia?

While preclinical experiments should go on to investigate these and other questions in more physiological models, such as xenografts of well-characterized human primary cultures of glioma, it is important to drive clinical studies on nutrient-restrictive approaches. Unfortunately, few well-structured studies have been performed thus far. Two pilot studies recently explored the application of KDs in patients with different solid tumors.^14,15 Concerning glioma, in 1995 Linda Nebeling described 2 pediatric patients with astrocytic tumors in whom a KD seemed to inhibit tumor growth.¹⁶ The ERGO study¹⁷ was the first prospective and registered clinical trial on KD in glioma. ERGO demonstrated that a KD can be safely applied to patients suffering from recurrent malignant glioma and that KD might increase the efficacy of anti-angiogenic therapies in a mouse xenograft model. However, ERGO also demonstrated that a calorically nonrestricted diet in glioma patients does not substantially lower blood glucose, most likely due to hepatic gluconeogenesis.

While additional studies on unrestricted KD in glioma patients are ongoing (NCTNCT02149459, NCT02046187, NCT02286167), preclinical results strongly suggest that caloric restriction might be necessary to inhibit tumor growth.⁵ Whereas long-lasting caloric restriction may be difficult to be implemented by patients and could be limited by weight loss, the concept of short-term fasting might be more applicable. For example, Longo and colleagues recently showed that transient fasting impaired tumor growth and decreased toxicity of conventional therapies.^18,19 We therefore recently set up the prospective and randomized ERGO2 trial to investigate transient carbohydrate and caloric restriction and fasting in patients undergoing radiotherapy for recurrent malignant glioma (NCT01754350). Other studies are investigating a calorie-restricted KD alone for refractory glioma (NCT01865162) or during first-line treatment with radiotherapy and temozolomide chemotherapy (NCT02302235). More sophisticated diets designed to reduce the burden of cyclic fasting that have shown beneficial effects primarily on longevity in mice may also be of interest in oncology.²⁰

Hopefully, these studies will help us to understand how restrictive therapies can be implemented and how they may be able to complement established therapies. As dietary interventions traditionally lead to polarizing debates—while neither supporters nor opponents have solid clinical data for their opinion—unbiased communication and careful conducting of clinical studies are necessary. Given the accumulating preclinical evidence for restrictive approaches, it is time to start these now.

Conflict of interest statement. JS has received a grant from Merck as well as honoraria for advisory board participation from Roche and Mundipharma and honoraria for lectures and/or travel grants from Roche, Boehringer, Bristol-Myers Squibb and Medac. JR has received honoraria for advisory board participation from Bristol-Myers Squibb.