The therapeutic ketogenic diet (KD) is an accepted therapy for a number of seizure disorders, and it continues to gain popularity for the treatment of malignant brain tumors in the adult and pediatric populations. Published data using pre-clinical models has demonstrated slowed tumor growth, enhanced survival, reductions in peritumoral edema, angiogenesis (formation of new blood vessels), inflammation, tumor hypoxia (reduced oxygen), the expression of a number of genes that regulate many tumor growth pathways (transcriptional activators), and enhancement of the antitumor immune response. Perhaps most importantly for patient care, they also showed it is an effective adjuvant treatment that potentiates the activity of radiation and chemotherapy. Furthermore, the increasing number of published case reports and anecdotal reports in social media are causing more cancer patients to approach their physicians with questions about the utility of this therapy.

Just as the anti-epileptic mechanisms of the KD have not been fully elucidated, the mechanisms behind the multifaceted effects of the KD on tumor cells is unclear. One effect is likely to be due to the reduction in blood glucose that results from a KD, since glucose is a preferred fuel for cancer cells. However, this is not sufficient to explain the variety of anti-tumor effects attributed to the KD in vivo and ketones in vitro. In fact, metabolic alterations resulting from increased ketones affect a wide variety of cellular characteristics which often differs in tumor cells and normal cells.

To understand the effects the ketogenic diet on malignant brain tumors we used a well established immune competent model of malignant brain tumors. Mice were surgically implanted intracranially with GL261-luc2 tumor cells and fed either a ketogenic diet (KD) or standard mouse diet (SD). Tumors were allowed to grow for approximately 3 weeks, then mice were humanely sacrificed and RNA was isolated from the tumors and from the opposite (nontumor containing) side of the brain. These RNA samples were used for gene expression analyses using a technique called RNA-Seq. In addition, we harvested RNA from brain samples from mice without implanted tumors that were maintained on a SD or KD to analyze gene expression differences in normal brain in the absence of tumor effects. Finally, we also implanted GL261-luc2 tumor cells grown in the laboratory with the addition of the ketone βhydroxybutyrate (BHB) to select for cells whose growth was not fully inhibited by BHB treatment. Gene expression changes in cells that were not fully inhibited by ketones may model what might occur in tumors that recur in patients that use a KD as part of their therapy.

The analysis of large RNA-Seq data sets such as what we have obtained from these experiments requires extensive statistical analysis, followed by validation of differential expression of specific genes and alterations in pathways implicated by these analyses. As expected, there were many genes that were differentially expressed in tumors versus the nontumor side of the brain, regardless of diet; however, there is a trend towards a higher number of differentially expressed genes in animals maintained on the KD. While this is also true for the tumors from the BHB-resistant cells, overall there are many more differentially expressed genes in tumors from the BHB resistant cells when compared to those from the parental GL261-luc2 cells. This was a surprising finding that we are investigating further. We are beginning with the few genes that were differentially expressed in both GL261-luc2 and GL261-luc2 BHB resistant tumors from mice fed a KD vs SD, followed by genes that may contribute to BHB resistance.

In addition to the analysis of tumors in mice fed a KD vs a SD we also compared the non-tumor containing sides of the brain from animals maintained on a KD vs a SD, and we compared these samples to those from normal brain from animals that were not implanted with tumors. We found a limited number of genes that were differentially expressed in the non-tumor containing side of brains from animals that had tumors, compared to animals that did not have tumors. This suggests that the presence of a tumor effects the non-tumor containing side of the brain. Again, even when analyzing the non-tumor containing side of the brain the number of differentially expressed genes was a bit higher in animals maintained on a KD. We were intrigued to find that there were also more differentially expressed genes in the non-tumor containing side of the brain from animals implanted with the BHB resistant cells compared to the parental GL261-luc2 cells.

We are currently validating the differential expression of specific genes identified by these analyses and working to put this data in the context of the anti-tumor effects seen with a KD. The gene expression analyses funded by this grant from Matthew’s Friends has provided a great deal of information that will continue to provide fertile ground for new discoveries regarding the mechanisms of the anti-tumor effects of a ketogenic diet. We are very grateful for their support.