The ketogenic diet is not effective in preclinical models of IDH1 wild-type and IDH1 mutant glioma

Infiltrative gliomas are the most common neoplasms arising in the brain, and remain largely incurable despite decades of research. A subset of these gliomas contains mutations in isocitrate dehydrogenase 1 (IDH1mut). This mutation disrupts cellular biochemistry, and IDH1mut gliomas are generally less aggressive than IDH1 wild-type (IDH1wt) gliomas. Some preclinical studies and clinical trials have suggested that a ketogenic diet (KD), characterized by low-carbohydrate and high-fat content, may be beneficial in slowing glioma progression. However, not all studies have shown promising results, and to date, no study has addressed whether IDH1mut gliomas might be more sensitive to KD. The aim of the current study was to compare the effects of KD in preclinical models of IDH1wt versus IDH1mut gliomas. In vitro, simulating KD by treatment with the ketone body β-hydroxybutyrate had no effect on the proliferation of patient-derived IDH1wt or IDH1mut glioma cells. Likewise, KD had no effect on the in vivo growth of these patient-derived gliomas. Furthermore, mice engrafted with Sleeping-Beauty transposase-engineered IDH1wt and IDH1mut glioma showed no difference in survival while on KD. These data suggest that IDH1mut gliomas are not more responsive to KD, and that clinical trials further exploring KD in this subset of glioma patients may not be warranted.

models of IDH1 wt versus IDH1 mut gliomas. In vitro, simulating KD by treatment with the ketone 23 body β-hydroxybutyrate had no effect on the proliferation of patient-derived IDH1 wt or IDH1 mut 24 glioma cells. Likewise, KD had no effect on the in vivo growth of these patient-derived gliomas.

25
Furthermore, mice engrafted with Sleeping-Beauty transposase-engineered IDH1 wt and IDH1 mut 26 glioma showed no difference in survival while on KD. These data suggest that IDH1 mut gliomas 27 are not more responsive to KD, and that clinical trials further exploring KD in this subset of glioma 28 patients may not be warranted. INTRODUCTION

31
Diffusely infiltrative gliomas strike over 17,000 people in the United States per year [1].

32
The vast majority of these tumors recur and progress, despite advancements in surgery, 33 chemotherapy, radiotherapy, and immunotherapy. In 20-39 year-olds, gliomas are the 2 nd most 34 common cause of cancer death in men, and are the 5 th most common cause in women [1]. The 35 most common type of primary brain cancer in adults is diffusely infiltrative glioma; the most 36 common subtype of infiltrative glioma, glioblastoma (GBM), is unfortunately also the most lethal.

37
Despite great advances in treating many other kinds of cancer, the median survival of GBM

103
To mimic the low glucose environment characteristic of physiological ketosis, a formulation of

133
Animals were placed on a weekly cyclic diet interchanging between SD and KD to control their 134 weights and more effectively maintain plasma ketone levels, as described elsewhere [15].

135
Animals were monitored bi-weekly for signs of morbidity such as weight loss, behavioral changes,

136
and hunched positions, and were euthanized when tumor size approached the recommended 137 limit of 2000 mm 3 , or when moribund, via CO 2 asphyxiation followed by cervical dislocation.

142
Differences between mean values of two groups were compared using two-sample t-test, 143 or between multiple groups by one-way analysis of variance (ANOVA) and post hoc Tukey's test; 144 P values less than 0.05 were considered significant. Log-rank tests compared survival between 145 groups. Graph generation and statistical analyses were performed with GraphPad Prism 5 146 (version 5.02, GraphPad Software, San Diego, CA).

148
First, we examined the effects of a ketogenic-like diet on cultured IDH1 wt and IDH1 mut 149 patient-derived cancer cell lines under the following conditions: (i) normal basal glucose (NG) of 150 4.5 g/L; (ii) low glucose (LG) of 1.0 g/L; (iii) NG with 10 mM β-hydroxybutyrate (BHB); (iv) LG with 151 BHB (Fig 1). Neither IDH1 wt GBM6 nor IDH1 mut HT1080 cells showed any difference in cell 152 number across the four culture conditions over time (Fig 1A and Fig 1D). Interestingly, IDH1 wt 153 GBM43 and IDH1 mut TB09 cells showed the greatest cell number when grown in LG +BHB ( Fig   154  1B and Fig 1C).

160
Next, we evaluated the ability of KD to affect the growth of patient-derived IDH1 wt and 161 IDH1 mut GBM cells in the flanks of immunocompromised mice (Fig 2).

171
Finally, to assess the effect of KD on orthotopic gliomas in the presence of an intact