Lessons learned from recent clinical trials of ketogenic diet therapies in adults

Abstract

Purpose of review

Although ketogenic diet therapies (KDTs) were first developed as a treatment for patients with epilepsy, their potential efficacy for a broader number of neurologic and nonneurologic disorders and conditions has been explored over the last 10–20 years. The most recent clinical trials of KDTs in adults have highlighted common methodological aspects that can either facilitate or thwart appropriate risk/benefit analyses, comparisons across studies, and reproducibility of findings in future studies.

Recent findings

Recent evidence suggests that KDTs not only improve seizure control, but also improve other neurologic conditions, including nonmotor Parkinson’s disease symptoms. Therapies targeting nutritional ketosis without comprehensive diet modification improve cognition and cerebral blood flow in Alzheimer’s disease patients. KDTs lower hemoglobin A1c levels and diabetes medication use in patients with Type 2 diabetes and mixed results have been observed when used for performance enhancement in athletes and healthy volunteers.

Summary

Clinical studies of KDTs show promise for a variety of clinical indications. Future studies should factor in high potential participant attrition rates and utilize consistent and standard reporting of diet type(s), compliance measures, and side-effects to enable the reproducibility and generalizability of study outcomes.

INTRODUCTION

The ketogenic diet is a high-fat, low-carbohydrate diet that induces ketone body production through fat metabolism with the goal of mimicking a fasting state, shifting the predominant caloric source from carbohydrate to fat [1,2]. The classic ketogenic diet is typically composed of a 4: 1 ratio (in grams) of fat to protein and carbohydrates. More ‘relaxed’ or modified variant forms of the ketogenic diet have emerged over the last 20 years to decrease rigidity and improve compliance. The modified Atkins diet (MAD) is typically composed of a net 10–20 g/day carbohydrate limit, which is equivalent to a ratio of 1–2: 1 of fat to protein and carbohydrates [3]. The low glycemic index treatment (LGIT) recommends 40–60 g daily of carbohydrates (carbs) with glycemic indices less than 50 and approximately 60% of dietary energy derived from fat and 20–30% from protein [4]. In addition, there is growing interest in therapeutic uses of ketogenic dietary supplements such as ketone esters and salts as well as medium-chain triglycerides (MCTs), commonly present in coconut and/or palm kernel oil, as alternative methods of achieving nutritional ketosis. Although ketogenic diet therapy (KDT) has been in use for nearly one century for the management of epilepsy, the last two decades have seen a surge in clinical interest in using KDTs in the fields of neurology, oncology, obesity, diabetes, and performance enhancement, among many others.

Recent clinical trials of ketogenic diets in adults

Recent evidence from clinical trials of KDTs in adults highlights the importance and variability of reporting of diet type used, study attrition rates, methods for assessing dietary compliance, and reported side-effects related to diet therapy. Accurate reporting will enable appropriate risk/benefit analyses, comparison across studies, and reproducibility of findings in future clinical trials. In example, the role of KDTs in oncologic clinical trials and case studies has been recently and comprehensively reviewed [5■■,6] with early evidence supporting the overall feasibility of varying ketogenic diets with or without caloric restriction, collective attrition rates of 50%, adherence measured by attainment of urinary and/or blood ketosis, and adverse effects, including fatigue, constipation, diarrhea, vomiting, and hyperuricemia reported most commonly. Additional randomized controlled studies on the effects of specific KDTs on tumor control and overall survival are warranted not only in patients with advanced cancer, but also those with earlier stages who may respond better to treatment. Highlighted clinical trials, particularly randomized studies (Table 1) [7■,8,9■,10,11], of KDTs in adults for other clinical indications published over the last 18 months are summarized below.

Table 1.

Highlighted recent RCTs investigating therapies targeting nutritional ketosis with comprehensive diet modification in adults

Disease	Reference	Study design	Pts (n)	KDT type	Duration (months)	Control treatment	Adherence measure	Study results	KDT attrition (reason)
Epilepsy	Kverneland et al. [7■]	RCT	75	MAD	3	Usual diet	Urine ketone level	No overall difference in responder rate between groups; more patients with 25–50% seizure reduction in diet group	13/37 (did not start diet, increased seizures)
	McDonald et al. [8]	RCT crossover	80	MAD + ketogenic formula supplement daily	1	MAD alone	Self-report of diet use and food records; urine and serum ketone level also measured	No significant difference in responder rate between groups, but trend for continued diet use in supplement arm	24/80 (did not start/stopped diet or supplement, change in seizure medication)
Parkinson’s disease	Phillips et al.[9■]	RCT	47	KD (< 16 g net carb/day)	2	Low-fat diet (< 42 g fat/day)	Serum ketone level (mean weekly ketones 1.15±0.59 mmol/l)	KD group with greater improvement in nonmotor symptoms	6/24 (tremor/rigidity, social reasons)
Performance enhancement	Greene et al. [10]	RCT crossover	14	< 50g carb/day	3	Usual diet (>250 carb/day)	Serum ketone level (0.4±0.2 mmol/l in diet group) and self-report of diet intake	Reduced total and lean body mass; no difference in fat mass or strength	2/14 (injury, scheduling conflict)
	Vargas et al. [11]	RCT	24	Hypercaloric KD (<42g carb/day)	2	Hypercaloric non-KD or usual diet	Urine ketone level	Reduced fat mass and visceral adipose tissue from baseline	2/11 (did not achieve ketosis)

carb, carbohydrate; g, grams; KD, ketogenic diet; KDT, ketogenic diet therapy; MAD, modified Atkins diet; NR, not reported; Obs, observational study; Pts, participants; RCT, randomized controlled trial.

Epilepsy

A Norwegian study [7■] randomized 75 adults with drug-resistant focal epilepsy to either 12 weeks of treatment with MAD (16 g carbs/day) vs. continuation of their habitual diet. Seventy-three percent (24/37) of patients in the MAD arm completed the 3-month study. Of the 13 patients in the MAD arm who withdrew, nine never started the diet and four were lost to follow-up because of intolerance or an increase in seizures. There was no significant difference between the diet and control groups in seizure frequency after the intervention (P = 0.21); however, more patients in the diet group achieved 25–50% reduction in seizures compared to controls (P = 0.03). To evaluate adherence, urinary ketosis was assessed using urine dipsticks with mean morning and evening levels of urine ketone production in the diet group of 2.9 mmol/l [95% confidence interval (CI) 1.6–4.1] and 5.9 mmol/l (95% CI 2.2–8.0), respectively. The most common reported adverse effects in the diet group were gastrointestinal symptoms that included nausea/vomiting, reflux, constipation, and diarrhea.

Our group performed a randomized 2-month cross-over controlled trial in 80 adults with drug-resistant epilepsy [8], with participants randomized to receive treatment with either MAD alone (20 g net carbs/day) or MAD supplemented with a 4:1 ratio ketogenic formula. Seventy percent (56/80) of patients completed the study. Of the 24 participants who did not complete the study, six never started the diet, seven stopped the diet and/or the supplement prematurely, six changed their seizure medications during the study period or did not consistently record seizure number, and the remainder were lost to follow-up. 47.5% of participants in the supplement arm were responders (≥ 50% reduction in seizures) and 37.5% in the diet alone arm, with no significant difference between groups (P = 0.498). However, there was a trend for higher continuation of MAD beyond the study period in the supplement arm (14/35–40%) compared to diet alone arm (7/36–19%; P = 0.072). Diet adherence was measured using patient self-reported continued diet use and diet-intake via food records. Levels of urinary and/or serum ketosis were also evaluated at follow-up. The most common adverse effects were constipation or diarrhea as well as a change or increase in seizure pattern/semiology.

Other neurologic disorders

A New Zealand study [9■] randomized 47 adults with Parkinson’s disease to either 8 weeks of a ketogenic diet (16 g net carbs and 152 g fat/day) or a low-fat diet (42 g fat and 246 g net carbs/ day). Eighteen of the 24 patients randomized to the ketogenic diet arm completed the study (25% attrition rate). Of the nine patients who withdrew from the study, three withdrew for diet-related difficulties (two in the ketogenic diet group). Although both diet groups showed improvements in motor and nonmotor symptoms on the International Parkinson’s and Movement Disorders Society UPDRS rating scale, patients in the ketogenic diet group showed greater improvements in nonmotor symptoms. As an indirect measure of diet compliance, blood ketones were measured with a ketone meter provided to patients with mean weekly blood ketone reading over the 8-week study in the ketogenic diet group 1.15±0.59mmol/l. The most common reported adverse effects were exacerbated tremor or rigidity in the ketogenic diet group.

Twenty adolescent or adult patients with relapsing-remitting multiple sclerosis at the University of Virginia, 95% of whom were overweight or obese, were enrolled in a 6-month single-arm study of the MAD (20 g carbs/day) [12]. Eighteen patients completed the 6-month study and two patients stopped the diet prematurely (10% attrition). A significant decrease in Expanded Disability Status Scale was observed at 6 months on the diet compared to baseline (P < 0.0001), primarily because of improvements in bowel/bladder and sensory functional systems. Fasting serologic leptin levels, a proinflammatory adipokine, were also significantly reduced from baseline at 3 months (P < 0.0001). Adherence was measured by home urinary ketosis and reported to study personnel using daily, dated pictures of test strips. Participants were considered ‘adherent’ if they demonstrated urinary ketosis at least 85% of the days of the 6-month study. Of the 18 patients who completed the study, 15 (75%) met this adherence criterion. The most common reported adverse effects were intermittent constipation, menstrual irregularities, and diarrhea.

A Turkish study [13] evaluated 350 patients with migraine without aura treated for 3 months with either LGIT or medical management. At 3 months of follow-up, 147 participants in the diet group were compared to 147 age-matched and sex-matched participants who had received medical management. It appears that 56 participants did not complete the 3-month study (16% attrition), although the authors do not mention what happened to these participants. The frequency and severity of headache attacks were measured with a visual analog scale (VAS) as well the mean number of attacks per month. Both the diet and medical management group showed reduced mean VAS scores and mean attack number at 3 months with no significant differences between groups, but only the medical management group had a significant reduction in VAS score at 1 month. It is unclear how dietary compliance was measured and adverse events were not reported.

Obesity and diabetes management

A Spanish study explored the effects of 2–3 months of a very low calorie ketogenic diet (600–800 kcal/day and < 50 g carbs/day from vegetables) to achieve a prespecified weight loss target followed by a 2-month low calorie diet (800–1500 kcal/ day) on food craving and psychological well being in 20 obese adults (12 women and 8 men) [14]. All 20 participants completed the study (0% attrition). The dietary-induced changes in body composition were associated with significant improvements in food craving, physical activity, sleepiness, and female sexual function. Adherence was assessed using daily blood ketone levels measured by a portable meter at home as well as at follow-up study visits, with levels averaging more than 1.0mmol/l during the period of maximum ketosis. No specific adverse effects were reported.

A multicenter randomized controlled feeding trial compared the effects of 20 weeks of a low-carbohydrate diet (20% daily intake; 105 g/ day) vs. a moderate-carbohydrate diet (40% daily intake; 205 g/day) vs. a high-carbohydrate diet (60% daily intake; 305 g/day) on energy expenditure during weight loss maintenance in 164 obese adults [15]. Daily protein intake was fixed at 20% of daily intake (102–104 g/day) and fat intake was 60, 40, and 20% in the low-carbohydrate, moderate-carbohydrate, and high-carbohydrate diet groups, respectively. Of 57 patients randomized to the low-carbohydrate diet, 43 completed the 20 week study (25% attrition). Total energy expenditure was significantly greater in the low-carbohydrate group (by 209 kcal/day) and moderate-carbohydrate group (by 91 kcal/day) compared to the high-carbohydrate group. Ghrelin and leptin levels were also significantly lower in the low-carbohydrate group. Adherence was assessed using a biomeasure of low-carbohydrate intake, 1,5-anhydroglucitol, as well as changes in triglyceride and high-density lipoprotein (HDL) cholesterol. Triglyceride levels increased with increasing carbohydrate content, whereas 1,5-anhydroglucitol levels and HDL levels decreased (P < 0.0001). Adverse events related to diet included food allergies, aversions, and intolerance in each diet group.

In adults with Type 2 diabetes, an open label, nonrandomized controlled 1-year study of either usual care or a continuous care intervention (CCI), including a sustained carbohydrate restricted diet (<30 g carbs/day) to induce nutritional ketosis, was performed [16■,17]. Of 262 participants in the CCI group, 218 completed the 1-year study (17% attrition). After 1 year, participants in the CCI group had significantly lower hemoglobin A1c levels (7.6±0.09 to 6.3±0.07%), weight (reduction of 13.8±0.71 kg from baseline), and reduced diabetes medication use [16■]. The CCI group also showed improvements in several cardiovascular disease risk factors, including apolipoprotein A1, HDL cholesterol, high-sensitivity C-reactive protein, and small low-density lipoprotein (LDL) particle number [17]. Adherence was assessed with serum ketone levels measured by a ketone meter and reported to study personnel electronically. Individualized nutrition recommendations were provided to participants based on biomarker results to achieve and sustain nutritional ketosis within 0.5–3 mmol/l. Mean ketone levels were 0.54±0.04 mmol/l at 70 days and 0.31±0.03 mmol/l at 1 year compared to 0.17±0.01 mmol/l at baseline. No adverse events were attributed to the CCI, although there was a significant increase in LDL cholesterol compared to the usual care group.

Performance enhancement

A Spanish randomized study evaluated the effect of an 8 week hypercaloric ketogenic diet (39 kcal/kg/day, <10% carbohydrates or ~42g/day, 70% fat), a hypercaloric nonketogenic diet (39 kcal/kg/day, 55% carbohydrates, 25% fat), or habitual diet (control group) on body compositing during resistance training in 24 healthy adult men [11]. Of 11 participants assigned to the ketogenic diet group, two did not achieve ketosis and were excluded from the analysis (18% attrition). Participants in the ketogenic diet group showed a significant reduction in fat mass and visceral adipose tissue from baseline, whereas participants in the nonketogenic diet group exhibited significant increases in total body weight and muscle mass from baseline. Adherence was assessed by weekly urine ketone measurements. No adverse effects were reported, although as no significant change in lean body mass was achieved with ketogenic diet use, it may not be an optimal strategy to build muscle mass using mechanical-tension focused resistance training.

An Australian group [10] used a randomized, cross-over design to compare the effect of 12 weeks of usual diet (>250 g carbs/day) to a low-carbohydrate ketogenic diet (<10% daily carbohydrate intake or <50 g carbs/day, 70% fat) on body mass and performance in 14 intermediate to elite competitive lifting adult athletes. Twelve participants completed the study and two dropped out because of scheduling conflicts or an injury (14% attrition). Patients in the diet group showed a significant reduction in total and lean body mass, but no difference in fat mass or strength performance compared to controls. Adherence was assessed using online self-report of diet intake using a smartphone application, MyFitness-Pal, and serum ketone level measured by a portable ketone meter at home. Mean ketone levels in the diet group were 0.4±0.2mmol/l, range 0.2–1.7mmol/l. No adverse effects were reported.

Clinical trials of therapies targeting nutritional ketosis without comprehensive diet modification

Recent studies have also evaluated the use of MCT supplementation in Alzheimer’s disease patients [18,19] with cognitive benefits observed with chronic therapy and enhanced cerebral blood flow observed in patients lacking an ε4 allele. Ketone salts, such as β-hydroxybutyrate, have been studied in short-term trials of performance enhancement with mixed results [20,21]. A randomized cross-over, controlled trial of virgin coconut oil in an isocaloric breakfast showed no acute change in postprandial energy metabolism and lipid levels within 4 h of administration to overweight women [22]. Finally, 2 weeks of exclusive ketogenic drink supplement use (4:1 ketogenic diet or 1.7:1 ketogenic diet) vs. a balanced nutrition drink in obese adults resulted in positive effects on body weight/mass and lipid profile [23]. Table 2 [18–23] further summarizes the study designs, sample size, adherence measures, and attrition rates observed in these studies.

Table 2.

Highlighted recent clinical studies investigating therapies targeting nutritional ketosis without comprehensive diet modification in adults

Disease	Reference	Study design	Pts (n)	KDT type	Duration (days)	Control treatment	Adherence measure (result)	Study results	Attrition (reason)
Alzheimer’s disease	Ota et al. [18]	RCT crossover	20	MCT-ketogenic formula AC-1202 (20 g MCT)	1	Isocaloric formula w/o MCT	Serum ketones (significant increase in MCT group compared to controls)	No difference in cognitive test scores between groups	0/20
	Ota et al. [18]	Obs	19	Daily MCT-ketogenic formula AC-1202 (20 g MCT)	84	N/A	Serum ketones (no change from baseline levels)	Significant improvement in logical memory test scores and digit-symbol coding from baseline	3/19 (diarrhea)
	Torosyan et al. [19]	RCT	18	Daily MCT caprylidene (40 g)	45	Placebo (two patients only)	No method reported (authors state pts adhered to daily supplementation)	Significant long-term elevation in regional cerebral blood flow in patients lacking an ε4 allele only (five patients)	4/14 (did not complete four PET scans)
Performance enhancement	Waldman et al. [20]	RCT crossover	15	Ingestion of βHB salts (11.38 g)	1	Placebo	Serum βHB levels (significant increase in βHB group – 0.53 vs. 0.21 mmol/l)	No difference in cognitive performance or high-intensity cycling	0/15
	Evans et al. [21]	RCT crossover	19	Ingestion of βHB salts (0.38 g/kg)	1	Placebo	Serum βHB levels (significant increase in βHB group – 0.28±0.13 mmol/l)	Elevated HR and RER during exercise; no effect on perceived exertion or muscular efficiency 13/19 (68%) experienced GI adverse effects	0/19
Obesity	Choi et al. [23]	RCT	46	Exclusive ketogenic drink use (either 4:1 or 1.7:1 ratio)	14	Balanced nutrition drink	Serum ketone level (significant increase in both ketogenic drink groups – 2.05±1.55 mmol/l for 4 : 1, 1.33±0.85 mmol/l for 1.7:1)	Decreased body weight in all groups; improved lipid profiles in 1.7: 1 group and reduced appetite in ketogenic drink groups	16/46 (poor compliance, accidental ingestion, and so on)
	Valente et al. [22]	RCT crossover	17	25 ml Virgin coconut oil (VCO)	1	25 ml extra-virgin olive oil	No method reported	No difference in energy metabolism or cardiometabolic risk markers; but less hunger suppression, satiety, and fullness with VCO	2/17 (personal reasons)

βHB, beta-hydroxybutyrate; g, grams; GI, gastrointestinal; HR, heart rate; KDT, ketogenic diet therapy; kg, kilogram of body mass; MAD, modified Atkins diet; Obs, observational study; Pts, participants; RCT, randomized controlled trial; RER, respiratory exchange quotient.

CONCLUSION

As KDTs become more widely used in a variety of clinical settings, despite a diverse spectrum of clinical indications and primary outcome measures, certain themes have emerged from published clinical trials in adults that may impact and guide future studies. First, a distinction has to be made between studies using low-carbohydrate diets and studies using ketogenic diets aiming to achieve nutritional ketosis. Those studies of comprehensive diet changes that limited carbohydrate intake to less than 50 g/day or less than 10% of caloric intake appeared to achieve average blood ketone readings of 0.4 mmol/l or higher. Meanwhile, studies of the MAD or more restrictive carbohydrate limits achieved blood ketones levels more consistently more than 0.5 mmol/l or a ketogenic ratio of at least 1 : 1 (fat: carbohydrates + protein) if dietary food records were used as a marker of adherence. Studies exploring the utility of a ketogenic diet should employ measures of adherence and report levels of ketosis achieved upon diet implementation. Second, attrition rates in clinical trials ranged from 0 to 35% in both shorter (≤3 months) and longer trials (≥ 6 months) with studies of more severe carbohydrate restriction demonstrating higher attrition rates. Thus, sample size determination for future clinical trials should factor in possible attrition rates of up to 35% to ensure adequate statistical power for outcome assessments. Finally, although no serious adverse effects have been reported with ketogenic diet therapies in recent studies, no mention is made of possible diet-related side-effects in some – so, it remains unclear if they did not occur or if they were not assessed – which makes generalizability across studies challenging. Participants in future clinical trials should be counseled on likely side-effects reported across trials, regardless of disease phenotype or clinical setting, including gastrointestinal complaints, and additional side-effects should be consistently reported as secondary outcomes.

KEY POINTS.

• Recent evidence suggests that ketogenic diets may play an impactful role not only in seizure management, but also for symptom management in Parkinson’s disease, multiple sclerosis, Type II Diabetes, and obesity.

• Therapies targeting nutritional ketosis without comprehensive diet modification can improve cognition and cerebral blood flow in Alzheimer’s disease patients and may have other benefits.

• Short-term trials of KDTs for performance enhancement have demonstrated mixed results.

• Clinical studies of KDTs should utilize consistent and standard reporting of diet type, compliance measures, and side-effects to enable the reproducibility and generalizability of study outcomes.