Survival outcomes of rectal and head and neck cancer patients receiving radio-(chemo)therapy with a ketogenic diet. Results from a controlled clinical study (KETOCOMP)

Background: High-fat, low carbohydrate ketogenic diets (KDs) have been proposed to target the cancer hallmark of high glycolytic metabolism and synergize with radio- and chemotherapy. We here report survival outcomes of rectal and head and neck cancer (HNC) patients who underwent a KD during radio-(chemo)therapy. Materials and Methods: Patients had participated in the controlled KETOCOMP study. The intervention group had consumed a KD during radiotherapy, while the control group had maintained their standard diet. All patients were prospectively followed until the occurrence of disease progression or death to evaluate overall (OS), progression-free (PFS) and locoregional recurrence-free survival (RFS) with the Kaplan-Meier method and by computing restricted mean survival times. In order to simulate a randomized controlled trial and meta-analysis, patients in the KD groups were matched to control patients with propensity score matching and data was pooled. Results: Median follow-up was 77.4 (range 12.1-107.9) months (HNC) and 72.6 (16.5-127.1) months (rectal cancer), respectively. HNC patients exhibited non-significantly longer OS and PFS and a trend for longer RFS times (p=0.079). These trends became more pronounced in the propensity score-matched sample (log-rank test p-values <0.10 for all three outcomes). In rectal cancer patients, there was no evidence for a benefit of the KD group. Analysis of the matched and pooled sample revealed a significantly longer restricted mean PFS time in the KD group (p=0.027) and a trend for longer RFS times (p=0.062). OS was also longer in the pooled KD group, but not significantly (p=0.29). Conclusions: These data indicate the potential for synergistic effects of a KD combined with radio-(chemo)therapy. While sample sizes could have prohibited the observed survival differences from becoming statistically significant, our data are useful to inform future studies or meta-analyses.