Effectiveness and Safety of a Novel Care Model for the Management of Type 2 Diabetes at 1 Year: An Open-Label, Non-Randomized, Controlled Study

Introduction Carbohydrate restriction markedly improves glycemic control in patients with type 2 diabetes (T2D) but necessitates prompt medication changes. Therefore, we assessed the effectiveness and safety of a novel care model providing continuous remote care with medication management based on biometric feedback combined with the metabolic approach of nutritional ketosis for T2D management. Methods We conducted an open-label, non-randomized, controlled, before-and-after 1-year study of this continuous care intervention (CCI) and usual care (UC). Primary outcomes were glycosylated hemoglobin (HbA1c), weight, and medication use. Secondary outcomes included fasting serum glucose and insulin, HOMA-IR, blood lipids and lipoproteins, liver and kidney function markers, and high-sensitivity C-reactive protein (hsCRP). Results 349 adults with T2D enrolled: CCI: n = 262 [mean (SD); 54 (8) years, 116.5 (25.9) kg, 40.4 (8.8) kg m2, 92% obese, 88% prescribed T2D medication]; UC: n = 87 (52 (10) years, 105.6 (22.15) kg, 36.72 (7.26) kg m2, 82% obese, 87% prescribed T2D medication]. 218 participants (83%) remained enrolled in the CCI at 1 year. Intention-to-treat analysis of the CCI (mean ± SE) revealed HbA1c declined from 59.6 ± 1.0 to 45.2 ± 0.8 mmol mol−1 (7.6 ± 0.09% to 6.3 ± 0.07%, P < 1.0 × 10−16), weight declined 13.8 ± 0.71 kg (P < 1.0 × 10−16), and T2D medication prescription other than metformin declined from 56.9 ± 3.1% to 29.7 ± 3.0% (P < 1.0 × 10−16). Insulin therapy was reduced or eliminated in 94% of users; sulfonylureas were entirely eliminated in the CCI. No adverse events were attributed to the CCI. Additional CCI 1-year effects were HOMA-IR − 55% (P = 3.2 × 10−5), hsCRP − 39% (P < 1.0 × 10−16), triglycerides − 24% (P < 1.0 × 10−16), HDL-cholesterol + 18% (P < 1.0 × 10−16), and LDL-cholesterol + 10% (P = 5.1 × 10−5); serum creatinine and liver enzymes (ALT, AST, and ALP) declined (P ≤ 0.0001), and apolipoprotein B was unchanged (P = 0.37). UC participants had no significant changes in biomarkers or T2D medication prescription at 1 year. Conclusions These results demonstrate that a novel metabolic and continuous remote care model can support adults with T2D to safely improve HbA1c, weight, and other biomarkers while reducing diabetes medication use. ClinicalTrials.gov Identifier NCT02519309. Funding Virta Health Corp. Electronic supplementary material The online version of this article (10.1007/s13300-018-0373-9) contains supplementary material, which is available to authorized users.

• Diabetes: Diagnosis of type-2 diabetes without end-organ failure • Urinary albumin > 1 g·dL -1 or > 10 g·L -1 • Impaired hepatic function (Bilirubin >2 mg·dL -1 or >34.2 µmol·L -1 , Albumin < 3.5 g·dL -1 or <35 g·L -1 ) • Cholelithiasis or biliary dysfunction • Cancer requiring treatment in the past 5 years, with the exception of non-melanoma skin cancer • Chronic infectious disease requiring ongoing treatment • Other chronic diseases or condition likely to limit lifespan to <6 years • Non-English speaking • Unable or unwilling to participate in group sessions (e.g., plans to relocate within the next year) or conform to a carbohydrate restricted diet lifestyle (e.g., food intolerances, religious or personal restrictions) • Weight loss of >10% in past 6 months • Currently pregnant or nursing, or planning to become pregnant during the study • Major psychiatric disorder (e.g., schizophrenia, bipolar disorder) currently uncontrolled • Excessive alcohol intake (acute or chronic) defined as average consumption of 3 or more alcohol-containing beverages daily or consumption of more than 14 alcoholic beverages per week Supplementary Materials B.

Methods -Medication Management
All medications changes were tracked in a database visible to the medical provider and health coach. Upon commencing dietary changes, insulin and sulfonylurea doses were reduced or eliminated to safely adjust for anticipated decreases in glucose concentrations based on the study physician's clinical judgment of the participant's most recent HbA1c, participant-reported last two weeks of blood glucose measurements, and other concurrent anti-diabetic medications. Similarly, prescription of SGLT-2 inhibitors were discontinued at onset of dietary changes.
[1] In the following days and weeks, participants reported BHB and glucoses one to three times per day via the app, based on medical provider instructions. Health coaches and medical providers monitored BHB and glucose, and the medical provider prescribed medications changes as necessary based on glucose and BHB trends and concurrent medication use with a goal of maintaining any reported glucose (fasting, pre-meal or post-meal) between 120 and 180 mg·dL -1 (6.66 and 9.99 mmol·L -1 ) until the medication could be eliminated.
During medication management, the primary focus was on preventing episodes of symptomatic hypoglycemia in patients. Any reported blood glucose <100 mg·dL -1 (<5.55 mmol·L -1 ) prompted insulin or sulfonylurea reduction of 10-50%. Medication eliminations typically occurred by first discontinuing sulfonylureas, followed by short-acting, and then long-acting insulin. Every patient prescribed short-acting insulin was instructed not to take any insulin if blood glucose was <120 mg·dL -1 (<6.66 mmol·L -1 ); for blood glucose ≥120 mg·dL -1 (≥6.66 mmol·L -1 ), the newly prescribed dose was reduced in 20 -50% increments in anticipation of reduction in dietary carbohydrates until the bolus or sliding scale was no longer needed.
Thiazolidinediones, DPP-4 inhibitors, and GLP-1 were discontinued after elimination of insulin and sulfonylureas and reported blood glucoses were consistently under 100 mg·dL -1 (5.55 mmol·L -1 ). Metformin, given its effectiveness, low cost, and tolerability [2,3], was generally not eliminated unless the participants reported side effects or requested discontinuation.
At any time during the study, reported blood glucoses consistently above 180 mg·dL -1 (9.99 mmol·L -1 ), generally, >3 times within a week and sustained for >1 week, prompted a medication increase or addition. If the participant was previously prescribed insulin, long-acting insulin was the first medication re-started. If the participant was previously prescribed a DPP-4 inhibitor, GLP-1, or a thiazolidinedione without insulin or sulfonylurea use, these were re-started. Two participants self-resumed taking their SGLT-2 inhibitors without physician direction. In these participants, BHB was monitored at least once daily, and no values of concern were reported. In all participants, blood glucose levels were monitored for trends up and down, which prompted the appropriate medication adjustments.
Fifty percent of CCI participants entered the trial with statin prescriptions.
Among completers, prescriptions were discontinued at patient request (n=18). In some cases, prescriptions were added (n=10) or increased (n=3) when deemed beneficial. Medication adjustments occurred both via remote care and at on-site clinic visits.

Methods -Education
Education provided participants an understanding of the core concepts and implementation instructions to enable initial achievement of nutritional ketosis and behavior change. Content included the pathophysiology of diabetes, practical management of carbohydrate restriction while consuming protein in moderation and increasing fat intake with a goal of achieving nutritional ketosis, how to assess and utilize BHB as a biofeedback mechanism and implementing of behavior change techniques. Additional resources were provided to participants to supplement their education, such as guides, blog posts, and recipes.

Methods -Laboratory Analysis
Blood analytes determined via standard procedures at a CLIA accredited laboratory at the time of draw included the following tests and panels: HbA1c, complete blood count (CBC), complete metabolic panel (CMP), highsensitive C-reactive protein (hsCRP), uric acid, thyroid stimulating hormone (TSH), free T4 (FT4), and lipids. HbA1c was determined from EDTA-treated whole blood using high performance liquid chromatography on Bio-Rad Variant II Turbo 2.0. Complete blood count with differential was determined from EDTA-treated whole blood with flow cytometry on a Beckman Coulter DxH80. Lithium-heparin treated plasma was utilized to determine the CMP, hsCRP, and uric acid with a Beckman Coulter AU680 and to determine TSH and FT4 via chemiluminescent detection on the Beckman Coulter DXI600.
Estimated glomerular filtration rate (eGFR) was calculated from serum creatinine, gender, race, and age [4]. 5.2% of all participants were African American.
Aliquots of serum were stored at -80°C and thawed once for later analysis of blood lipids, insulin, C-peptide, and BHB by a CLIA accredited laboratory.   Difference between means or percentages ± 1 standard error of the difference. Significant baseline difference between means or percentages at 0.05>P≥0.01 (*); 0.01>P≥0.001 ( †); 0.001>P≥0.0001 ( ‡); and P<0.0001 ( §).  a Imputed values based on 700 iterations from multivariate normal regression. b Mean differences ± one standard error. Sample sizes, means, and significance levels refer to intervention subjects with baseline, 70-day, and one-year measurements for completers. Significance levels for completers refer to one-sample t-test. Untransformed triglyceride and hsC-reactive protein values are presented, however, their statistical significances were based on their log-transformed values. c Significance levels refer to two-sample t-test of the differences are designated for 0.05>P≥0.01 (*); 0.01>P≥0.001 ( †); 0.001>P≥0.0001 ( ‡); and P<0.0001 ( §). d A significance level of P<0.0017 ensures overall simultaneous significance of P≤0.05 over the 30 variables using Bonferroni correction.