In this pilot study, we evaluated the safety (occurrence of side effects), feasibility (following the dietary guidelines for advanced cancer patients) and impact (on body weight, laboratory parameters, quality of life) of a LowCarb/HighFat (LCHF) diet very low in carbohydrates (CHO) and rich in fat and protein on a heterogeneous group of advanced cancer patients. Announcing the study provoked a broad interest of media and the hospital was overwhelmed with enquiries from cancer patients. This reflects the enormous interest of cancer patients in complementary and alternative methods of improving their situation via change of lifestyle and, especially, their diet [41, 42]. However, most of the patients interested were either end-stage (hoping for a "miraculous cancer cure") or disease free after primary therapy, intending to use the diet for prevention of relapse. Both groups were not eligible, thus reducing the number of patients fulfilling the requirements to very few. It should be noted in this respect, that due to the heterogeneity of the patients included and the nature of this diet with normal food, this study was a series of cases and neither randomized nor blinded.
As described in previous studies , we found an LCHF treatment in adult patients to be only slightly feasible. Based on our observations, an LCHF diet is not an option for all patients with advanced cancer, since the associated changes in eating habits (e.g. waiving soft drinks and beer) are not acceptable for some of them. For other patients however, an LCHF diet might be an option for increasing their quality of life. The latter belong to the group of patients who want to actively influence the course of the disease (by change of lifestyle) and who are in a phase of their disease where cooking and eating are not hampered. Although there was a worsening in some parameters of the quality of life assessment, reflecting the very advanced situation of our patients, we found an improvement in emotional functioning and insomnia, even though the course of the disease in our patients was progressive or in the best case, stable. Certainly, we cannot exclude the possibility of a placebo effect caused by a) the intensive consultation and briefing of the patients and b) the chance for patients to actively participate in their therapy. Further, anecdotic evidence links the presence of ketone bodies to a mild euphoria  which was assumed to be caused by one of the ketone bodies, beta-hydroxybutyrate. Thus, this metabolic state of ketosis, which was reached by six of our patients, could be the reason for the improvement in our patients, too.
Diets high in fat (up to 90%) and very low in CHO, inducing a stable ketosis, have been used for a long time to treat epilepsy and adiposity. The most frequent side effects reported so far are constipation, vomiting, lack of energy and hunger . These symptoms were reported by our patients slightly more often within the first 4 weeks of the diet. However, none of them complained of hunger, and nausea and vomiting were very rare. We could also not determine from our data, whether the observed increase in fatigue was caused by the diet or by the progress of disease. In this context we can also report, that patients outside the study and following the diet during standard therapies (chemo-, radiotherapy), regularly report an increase in energy and condition (not shown).
The time frame of the study was selected according to common duration of studies dealing with the application of ketogenic diets in epilepsy patients [20, 40] and based on the short survival perspective of the patients enrolled. Further, the Nebeling study showed that the effects of a ketogenic diet were not visible until after 8 weeks . Since our patient group was small and diverse, we cannot comment on the influence of the LCHF diet on the course of disease. The influence of the diet on specific tumor entities or on tumors with different molecular characteristics concerning glucose metabolism must be evaluated in further studies.
The present study has several limitations. Only few patients fulfilled the requirements and all of them were in a very advanced stage of disease, as reflected by the two early death cases and the progress of disease in five cases which made it impossible for those patients to follow the diet for the whole time. Since we aimed to test the acceptability and compatibility of an LCHF diet in advanced cancer patients, we did not select patients according to their tumors. Therefore, the group was very inhomogeneous. Further, the majority of the study participants were not from our own hospital, but scattered all over Germany and blood samples and laboratory parameters had to be provided by their family doctors or local oncologists.
Compared to KD regimens for the treatment of epilepsy or obesity that reduce the amount of CHO allowed to at least 10-25 g per day, our LCHF protocol allowed up to 70 g CHO/day. This larger amount was allowed due to several considerations: 1) The daily glucose production rate in healthy patients was determined to be around 3.6 µmol/kg/min (corresponding to 0.933 g/kg/day) with a very stable rate of gluconeogenesis of 2.6 µmol/kg/min (corresponding to 0.67 g/kg/day) and a variable rate of glycogenolysis. The latter depends on the amount of glucose/protein and, if too much CHO was taken up, glucose was stored as glycogen . 2) Increased rates of gluconeogenesis, which burns the body's lean mass and harms the patient, have been documented in patients with malignant disease . Thus, advanced cancer patients hypothetically may tolerate a little more CHO in their diet without leaving the metabolic state ketosis. 3) The larger amounts of CHO allow patients to add yoghurt and some vegetables/fruits, which although containing milk or fruit sugar, are also promoted to be beneficial to cancer patients  and - based on the popularity of these books - very welcomed by our patients. These facts in mind, allowing a little more CHO in the diet should not severely influence the ketogenic effect and indeed, 6 of the 11 patients on the diet for at least 6 weeks reached ketosis. More CHO facilitates the selection and compilation of food and thus increases the compliance of the patients. From data available, we cannot determine the reasons for the failure of the remaining patients to reach a stable ketosis. It could be speculated in this respect, that either the amount of CHO allowed was over the individual ketogenetic limit, or the patients consumed additional CHO without documenting it. The method for analyzing urinary ketones does not seem to be the cause of negative results, since this method shows values that correlate highly with blood ketone values in a preliminary test series (not shown) and was already described in literature to show a good correlation to blood ketones in long time ketosis . However, to help achieve a stable ketosis state in further trials, a step-by step induction of ketosis as suggested by Atkins  may be necessary, with a CHO content of the diet customized to the preservation of ketosis in each individual patient.
The significant reduction of body weight (and BMI) observed in all patients could reflect the typical effect observed in the early phase of low-carbohydrate diets - usually not restricted in calories. Here, the maximum weight loss was normally observed within the first 6 months of the diet [25, 48, 49]. In contrast to the weight loss observed in our group and described in general for ketogenic diets, a non-ketogenic high-fat diet (mixture of normal meals with an additional fat-enriched artificial liquid diet) supported the maintenance of body weight in patients with gastrointestinal tumors . However, those patients were not "end stage" and the diet was followed during chemotherapy. Yet, based on their findings, it could be speculated that the application of an LCHF diet in an earlier phase of tumor disease might be of greater benefit for the patient.
Among the study patients and treating oncologists, many concerns were expressed that a diet very rich in total fat could adversely affect blood lipid levels or immune status. In accordance to data published in several extended studies applying low-carb diets to obese patients to lose weight [25, 48, 49], the cholesterol and especially LDL/HDL values of our patients significantly improved. The exception to this pattern is the triglyceride concentration. Other than described in the cases of ketogenic diets for the reduction of body weight , the levels of triglycerides in our patient's sera increased on the diet, albeit not to a significant level, and still in normal range. This effect was also observed by Mosek  and thus may reflect a normalization of the triglyceride level in non-overweight patients. However, we have not investigated subpopulations of immune cells, other than described in Breitkreutz  who observed a significant decrease in total leucocyte count in the group of patients on the fat rich diet. The leucocyte count in our patients significantly increased to the better. This difference may be due to different settings and the accompanying chemotherapy of the patients in the former study.
Since several studies have shown that supplementation with Omega-3 fatty acids (FA) benefits patients with advanced cancer and weight loss [50–52], the vegetable oil used in the oil-protein shake taken by the patients two times a day was especially enriched in Omega-3 FA. Further, the patients were encouraged to snack on nuts and seeds rich in Omega-3 FA such as flaxseed, hempseed and walnuts. However, the study was too short and the patients presumably too advanced in their disease to observe a beneficial effect on tumor growth as described in animal models [31, 32, 53].
When we started the study in 2007, except for two preliminary reports [26, 54], no protocol was available on how to perform an LCHF or ketogenic diet study with cancer patients. Since then, a study protocol was published by Fine et al. , and four clinical trials were registered in the clinical trials database . However, no patient's data resulting from these studies have been published so far. Thus, it will be very interesting to see if the slightly different nutritional settings of LCHF or ketogenic diets will benefit the study patients as well.