Understanding the Paradoxical Rise in Blood Glucose in the Context of a Low-carbohydrate Diet

Using a randomized crossover design with counterbalanced treatment conditions, participants were enrolled in the study for the duration of 4 weeks. Baseline testing took place during the first week and the three intervention conditions took place from weeks 2-4. Participants completed all three intervention arms with a 6- to 10-day washout pe-riod between each intervention session. The order of conditions was randomly deter-mined (randomizer.org) and assigned to participant numbers prior to the start of the study. Participant numbers were given sequentially based on the order in which they signed the consent form and officially enrolled in the study. Before each intervention session, participants were screened for contraindications to ensure continued eligibility. Approval from the university's Institutional Review Board was obtained prior to initiating any aspect of this study.

2.1 Participants

Participants were between 18 and 65 years of age and were weight stable (± 3% body weight) for the past 3 months. Additionally, participants must have been adhering to a low-carbohydrate (50g carbohydrates or less per day) diet at least 2 weeks prior to and during the duration of the one-month study. Requirements for inclusion were vetted through the initial qualification screening and further confirmed at the initial screening appointment. Individuals interested in the study were sent a qualification survey (Qualtrics.com) before enrollment to determine participation eligibility. Potential par-ticipants were excluded if they did not provide written consent for participation or if they met any of the following criteria:

1. Diagnosed with chronic disease(s) including cardiac disease, cancer, or liver disease
2. Were pregnant or lactating
3. Were taking medications including anticoagulants and those that may alter me-tabolism (i.e.- insulin, metformin, and amphetamine-based ADHD medications)
4. Had allergies to latex, cow's milk/milk products, olives, chlorhexidine, or dextrose.
5. Have had more than 5 X-rays within the past year

2.2 Measurements Blood pressure and heart rate were taken using an automated sphygmomanometer (Omron Inc., Model HEM-907XL) at each visit prior to phlebotomy to ensure safety. If the systolic pressure was less than 100mmHg or if the diastolic pressure was below 60mmHg, phlebotomy was not performed.

2.2.1 Anthropometric and body composition measures Body weight and height were measured for all participants at the beginning of each session. Weight was measured using a digital scale (Seca, Hamburg, Germany) accurate to ± 0.1 kg with participants dressed in athletic shorts and a t-shirt and with shoes re-moved. Height was measured with shoes removed by a stadiometer accurate to ± 0.1 cm (Seca, Hamburg, Germany). Body mass index (BMI) was calculated as weight (in kilo-grams) divided by the square of height (in meters). A GE iDXA (GE, Fairfield, CT) was used to assess fat-free mass, fat mass, lean mass, percent body fat and visceral adipose tissue[8]. Calibration of the dual-energy X-ray absorptiometer (DXA) scanner took place at the beginning of each testing day using a manufacturer-provided calibration block. Scans were analyzed using Encore software version 17.

2.2.2 Plasma hormone concentrations The Human Metabolic Hormone Magnetic Bead Panel multiplex kit (Sigma-Aldrich, Inc., St. Louis, MO; Catalog # HMHEMAG-34K) was utilized to quantify the plasma concentrations of 13 metabolic hormones including amylin, ghrelin, glucagon, gluca-gon-like peptide 1 (GLP-1), glucose-dependent insulinotropic polypeptide (GIP), inter-leukin 6 (IL-6), insulin, c-peptide, leptin, monocyte chemoattractant protein-1 (MCP-1), pancreatic polypeptide (PP), peptide YY (PYY), and tumor necrosis factor alpha (TNF-α). Venipuncture took place in the human performance laboratory by a trained phlebotomist using standard phlebotomy procedures, including sterile technique, to ensure that risks to the participants were minimized. A single 4 mL vacuum-sealed tube prepared with Ethylenediaminetetraacetic acid (EDTA) was taken from each participant at or near the median cubital vein. Each 4 ml tube was inverted to allow for mixture with the EDTA and then samples were centrifuged for 15 minutes at 1500 rotations per minute within 10 minutes of collection, after which the plasma was aliquoted and placed into cryovials. Plasma samples were then stored at -80°F for later analysis.

2.2.3 Continuous glucose monitoring Interstitial subcutaneous glucose concentrations were measured using the Freestyle Libre 2 continuous glucose monitor (CGM) (Abbott Laboratories). The interstitial glucose concentration closely reflects the intravascular glucose concentration, making this a useful tool to measure the glycemic response to various stimuli[9]. Each CGM was inserted on the back of the non-dominant upper arm and given at least 60 minutes to calibrate, per manufacturer specifications, to ensure sensor accuracy[10]. This sensor recorded the interstitial glucose concentrations every 15 minutes and these data were downloaded to the participants' personal smartphone using the Freestyle Libre mobile application. These data were then uploaded to online cloud storage for later analysis.

2.3 Procedures Once eligibility was determined and informed consent was signed, participants were sent a link provided by Levels Heath, Inc. to order a one-month supply (two units) of continuous glucose monitors (CGM). Once the CGM units arrived in the mail, subjects scheduled their baseline assessment visit. For this visit and all intervention visits, subjects arrived at the human performance laboratory having fasted 8-12 hours, main-tained normal sleep patterns, and refrained from participating in strenuous exercise for the past 24 hours. Adherence to the pre-test day protocols were gauged at the beginning of each session. If the pre-test protocols had not been followed, the participant was re-scheduled. DXA scan was completed to measure the body composition parameters of interest. Participants were scanned wearing loose clothing with no jewelry or items in pockets and shoes off. Following the DXA scan, subjects underwent a single blood draw (4 mL) and a continuous glucose monitor was placed and activated according to manu-facturer recommendations.

2.3.1 Intervention sessions Following the baseline assessment, study subjects were scheduled for three inter-vention visits, spaced 6-10 days apart. At each intervention visit, subjects were screened for protocol adherence and one of three dietary conditions was initiated. The dietary conditions included a carbohydrate beverage, a protein beverage, and a fat beverage. Condition order was assigned randomly prior to the intervention visits for each partic-ipant. The carbohydrate beverage consisted of 300 kcal (88 g) of dextrose powder dis-solved in 12 oz of water. The protein beverage consisted of 300 kcal (45 g) of unflavored whey powder in water. The fat beverage consisted of 300 kcal (1.25 oz) of olive oil sus-pended in 12 oz of water.

At each session, venous blood samples were be taken at times 0 minutes, 30 minutes, 60 minutes, 90 minutes, and 120 minutes. Immediately following the first blood draw (0 minutes), the participant consumed the entire assigned beverage within 5 minutes. Between draws, participants were allowed to rest, either sitting or slowly moving about the room, but were asked to refrain from eating or excessive movement. At the conclusion of each visit, the participant was excused from the laboratory and returned on their as-signed days to complete the remaining conditions of the study.

2.4 Statistical Techniques An a priori sample size was estimated using a β-1 of 0.8, an alpha of 0.05, and moderate effect size (d=0.66) to detect a 30% difference in glucose area under the curve between dietary conditions. From these calculations, it was determined that a total of 21 participants were needed for adequate statistical power. Twenty-four participants were ultimately recruited anticipating that some individuals would be lost to dropout.

The data were described using numbers, percentages, means, and standard deviations. Prior to analysis, each variable was assessed for normality. Mixed-effects models were used to evaluate postprandial hormonal responses using the PROC MIXED procedure. Each hormone (glucagon, insulin, C-peptide, amylin, ghrelin, GIP, GLP-1, pancreatic polypeptide, PYY, TNF-α, IL-6, MCP-1, and leptin) was analyzed separately with condition (carbohydrate, fat, protein), time, and their interaction as fixed effects, and participant as a random effect. Sex and body mass index were included as covariates. Repeated measures across time were modeled using several covariance structures, in-cluding compound symmetry (CS), unstructured, and first-order autoregressive. Model fit indices (AIC, AICC, and BIC) were compared across structures, and CS provided the best overall fit; therefore, CS was used for all reported analyses. Significant main and inter-active effects were further examined using the LSMEANS procedure for pairwise com-parisons. Data were evaluated using the statistical software package (Base SAS 9.4; SAS Institute, Inc., 2016).