There was a resurgence of interest in very low calorie ketogenic diets for weight loss in the 1970's, followed closely by the complications (including sudden death) associated with the Liquid Protein diet popularized in 1976. However, the fatigue and apparent cardiac dysfunction caused by this collagen-based fad diet stood in stark contrast to the published experience of arctic explorers such as Schwatka and Stefansson. In addition, physicians who monitored patients following very low calorie diets observed wide variations between the exercise-tolerance of these individuals.
Given that the elegant research on the metabolism of total fasting by Dr. George Cahill and colleagues had demonstrated that full adaptation of nitrogen, fat, and carbohydrate metabolism required a number of weeks [12
], it seemed reasonable to hypothesize that exercise tolerance would take more than a week to recover after removal of carbohydrate from the diet. This view was supported by the subsequent discovery of the prescient adaptation quote from Schwatka's diary [6
] noted above.
To test this hypothesis, the current author (under the mentorship of Drs. Ethan Sims and Edward Horton at the University of Vermont) undertook a study of subjects given a very low calorie ketogenic diet for 6 weeks in a metabolic research ward [13
]. The protein for this diet, along with a modicum of inherent fat, was provided by lean meat, fish, and poultry providing 1.2 grams of protein per kg of reference ("ideal") body weight daily. In addition, mindful that the natriuresis of fasting could reduce circulating blood volume and cause secondary renal potassium wasting, the subjects were prescribed 3 grams of supplemental sodium as bouillion and 25 mEq (1 g) of potassium as bicarbonate daily.
Treadmill performance testing of these subjects included determinations of peak aerobic power (VO2max) after a 2-week weight maintenance baseline diet, and again after 6 weeks of the ketogenic weight loss diet. Endurance time to exhaustion was quantitated at 75% of the baseline VO2max. This endurance test was repeated again after one week of weight loss and finally after 6 weeks of weight loss. Other than these tests, the subjects did no training exercise during their participation in this study. To compensate for the fact that the average subject had lost over10 kg, the final endurance treadmill test was performed with the subject carrying a backpack equivalent in weight to the amount lost.
The energy expenditure data (expressed as oxygen consumption) and exercise times across this 8-week inpatient study are shown in Table . That these subjects'peak aerobic power did not decline despite 6 weeks of a carbohydrate-free, severely hypocaloric diet implies that the protein and mineral contents of the diet were adequate to preserve functional tissue. As can be noted, endurance time to exhaustion was reduced after one week of the ketogenic diet, but it was significantly increased over the baseline value by the 6-week time point. However the interpretation of this endurance test is confounded by the fact that the oxygen cost (ie, energy cost) of the treadmill exercise had significantly decreased following the weight loss, and this occurred despite the subjects being made to carry a backpack loaded to bring them back to their initial exercise test weight.
Exercise parameters of Vermont study 
This question of improved efficiency notwithstanding, it is clear that our subjects experienced a delayed adaptation to the ketogenic diet, having reduced endurance performance after one week followed by a recovery to or above baseline in the period between one and six weeks. Given the reduced energy cost of the exercise despite the backpack, the extent of this adaptation cannot be determined from this study. To explain this improved exercise efficiency, we can speculate that humans are more efficient carrying weight in a modern backpack than under their skin as excess body fat. It is also possible that these untrained subjects became more comfortable with prolonged treadmill walking by their third test, and therefore improving their overall efficiency.
Given the uncertainties of this study caused by the subject's weight loss and potential for improved technique with multiple tests, the current author undertook a second study under the mentorship of Dr. Bruce Bistrian at MIT in Cambridge MA [14
]. The diet employed in this followup study was patterned after that consumed by Stefansson during his year in the Bellevue study (and thus presumably close to that traditionally consumed by the Inuit) with the intention that the subjects would be in ketosis without weight loss.
This second study utilized competitive bicycle racers as subjects, confined to a metabolic ward for 5 weeks. In the first week, subjects ate a weight maintenance (eucaloric) diet providing 67% of non-protein energy as carbohydrate, during which time baseline performance studies were performed. This was followed by 4 weeks of a eucaloric ketogenic diet (EKD) providing 83% of energy as fat, 15% as protein, and less than 3% as carbohydrate. The meat, fish, and poultry that provided this diets protein, also provided 1.5 g/d of potassium and was prepared to contain 2 g/d of sodium. These inherent minerals were supplemented daily with an additional 1 g of potassium as bicarbonate, 3 grams of sodium as bouillon, 600 mg of calcium, 300 mg of magnesium, and a standard multivitamin.
The bicyclist subjects of this study noted a modest decline in their energy level while on training rides during the first week of the Inuit diet, after which subjective performance was reasonably restored except for their sprint capability, which remained constrained during the period of carbohydrate restriction. On average, subjects lost 0.7 kg in the first week of the EKD, after which their weight remained stable. Total body potassium (by 40K counting) revealed a 2% reduction in the first 2 weeks (commensurate with the muscle glycogen depletion documented by biopsy), after which it remained stable in the 4th week of the EKD. These results are consistent with the observed reduction in body glycogen stores but otherwise excellent preservation of lean body mass during the EKD.
The results of physical performance testing are presented in Table . What is remarkable about these data is the lack of change in aerobic performance parameters across the 4-week adaptation period of the EKD. The endurance exercise test on the cycle ergometer was performed at 65% of VO2max, which translates in these highly trained athletes into a rate of energy expenditure of 960 kcal/hr. At this high level of energy expenditure, it is notable that the second test was performed at a mean respiratory quotient of 0.72, indicating that virtually all of the substrate for this high energy output was coming from fat. This is consistent with measures before and after exercise of muscle glycogen and blood glucose oxidation (data not shown), which revealed marked reductions in the use of these carbohydrate-derived substrates after adaptation to the EKD.
Exercise parameters of MIT EKD study 
Examining the results of these two ketogenic diet performance studies together indicates that both groups experienced a lag in performance across the first week or two of carbohydrate restriction, after which both peak aerobic power and sub-maximal (60–70% of VO2max) endurance performance were fully restored. In both studies, one with untrained subjects and the other with highly trained athletes who maintained their training throughout the study, there was no loss of VO2max despite the virtual absence of dietary carbohydrate for 4–6 weeks. This whole-body measure of oxidative metabolism could not be maintained unless there was excellent preservation of the full complement of functional tissues including skeletal muscle (and mitochondrial) mass, circulating red cell mass, and cardiopulmonary functions.
The possibility raised by the first study of improved endurance time after keto-adaptation was not substantiated by the second study employing highly trained athletes without the complicating variable of major weight loss. It is thus likely that the increased endurance time in the Vermont study was due to improved efficiency (ie, less hobbling from a backpack than from an equal weight of internal body fat) and/or improved acclimation to the endurance test procedure. Such acclimation would not be expected in the second study, as the highly trained bicycle racers were well conditioned to the stationary ergometer at the start of the study. It is also worth noting that the bicycle racers remained weight stable (excepting the half kilogram of reduced muscle glycogen) across the 4 weeks of the EKD, which was equi-caloric with the baseline diet. Although 4 weeks is a relatively short period to assess small differences in energy efficiency between diets, this observation implies that there was no great reduction in the efficiency of energy metabolism after keto-adaptation.
As a final note in this section, neither the Vermont study nor the MIT study has been refuted in the 2 decades since their publication. Understandably given the expense of human metabolic ward studies and the orthogonal conclusions of these two studies, neither study has been corroborated by a similar human study. However two subsequent animal studies examining physical performance after keto-adaptation have yielded results consistent with those presented above [16