Clinical Therapeutics/New Technology— Glucose Monitoring and Sensing 2348‑Pub

Home , Diabetes management, Gemigliptin, Ipragliflozin, Luseogliflozin, Teneligliptin, Tofogliflozin

CLINICAL THERAPEUTICS/NEW TECHNOLOGY—GLUCOSECATEGORY MONITORING AND SENSING

CLINICAL THERAPEUTICS/NEW TECHNOLOGY— Figure. GLUCOSE MONITORING AND SENSING

2348‑PUB Impact of CME on Improving Understanding of Advances in Glucose Data Interpretation AMY LARKIN, MICHAEL LACOUTURE, ANNE LE, New York, NY A substantial advance in the field of glucose monitoring is the development of CGM devices. We sought to determine if online continuing medical education (CME) could improve the clinical knowledge and competence of diabetologists/endocrinologists (D/Es) and nurses regarding advances in glucose data interpretation. A CME activity was developed as an online video discussion between 2 experts. The effects of education were assessed using a 4-question linked pre-/post-assessment study design, McNemar’s chi-squared test, and Cramer’s V for effect size. Baseline knowledge was higher among D/E compared to nurses: · 63% of D/Es compared to 20% of nurses recognized similarities in glucose monitoring devices. 2350‑PUB · 47% of D/Es compared to 22% of nurses recognized clinical applica- Efficacy of PHR Integrated with EHR and Self-Monitoring Devices tion of glucose data. on Self-Care in Patients with Type 2 Diabetes SATOSHI TANIGUCHI, JIN TEMMA, AKIO KURODA, TORU HORIE, HIROYASU · 55% of D/Es compared to 36% of nurses recognized an effective MORI, REIKO SUZUKI, YAYOI ASANO, MICHIKO ARAKI, YU TAMAKI, MUNEHIDE strategy for patient engagement using glucose data. MATSUHISA, Tokushima, Japan Significant overall improvements (P < .05) were seen for both D/Es (n = 137; Background: Digital Personal Health Records (PHR) with tele-medicine medium effect V= 0.171) and nurses (n = 763; small-medium effect V= 0.15). have been proved to be effective in treating diabetes. However, simply digi- tizing logbooks (improving recording, reviewing and sharing capability) may Therapeutics

· 17% more D/Es and 24% more nurses correctly recognized A1c pro- Clinical Diabetes/

vides information about glucose exposure. be enough to increase adherence. PUBLISHED ONLY Objective: To develop and evaluate the efficacy of a PHR integrated with · 24% more D/Es and 8% more nurses correctly identified the ability to Electronic Health Record (EHR) and self-monitoring devices. use glucose monitoring data for advancing therapy. Method: A 3-month multi-center randomized trial was performed to type 2 · 20% more D/Es and 17% more nurses selected an effective strategy diabetes patients. Intervention group used the developed PHR “e-DM Diary,” for engaging a patient in their diabetes management plan using glu- with BG monitor, digital scale, BP monitor and activity monitor. Control group cose data. used self-monitoring devices with paper logbooks issued by Japan Asso- Additional education needed: ciation for Diabetes Education and Care, a standard protocol in Japan. The e-DM Diary or paper logbook was referenced at doctor visits. There were no · 36% of D/Es and 54% of nurses failed to recognize similarities in other modifications to treatment. glucose monitoring devices. Overview of e-DM Diary: e-DM Diary is accessed by smartphone or PC. Self- · 31% of D/Es and 71% of nurses failed to recognize information pro- monitoring results are automatically uploaded through patient’s smartphone vided by measuring A1c. via Bluetooth. Patient’s EHR, including test results, treatment goals, and pre- · 53% of D/Es and 63% of nurses failed to recognize clinical applica- scription information were provided automatically. Patient’s status of diabetes tion of glucose data. complications, self-monitoring goals and results, were presented in graphs. This study demonstrates the success of a targeted educational interven- Results: 28 participated were enrolled in this study (intervention=15/con- tion on improving knowledge and competence of D/Es and nurses regarding trol=13, age avg. 54 y.o., HbA1c 6.9%). Summary of Diabetes Self-Care Activ- clinical application glucose data interpretation advances. Baseline knowl- ities Measure, glycemic control and body weight did not change before and edge was higher among D/Es. Additional education on advances in glucose after the study period in both groups, and there was no difference between data is needed for both groups. groups. Patient’s age, e-Health literacy did not influence the results. In the intervention group, 12 of 15 patients answered that e-DM Diary was Supported By: Abbott Diabetes Care effective in reducing the “hassle” of self-monitoring. There were 3 cases were the patient’s understanding of diabetes complication improved. 2349‑PUB Conclusion: e-DM Diary’s capability to reduce hassle of self-care was Blood Glucose Telemonitoring at Retail Pharmacies in China: The recognized, and self-care supporting capability was par to paper logbooks. Alternative Sites for Community Diabetic Care Supported By: Japan Ministry of Internal Affairs and Communications WENHAO QU, YINGJIE LI, YING CHEN, ZHEN WANG, KAI LIU, Shanghai, China Objective: To implement a digital solution at retail pharmacies to improve community diabetic care. 2351‑PUB Methods: We developed a digital solution that consists of bluetooth glucose meter, an App and a cloud database for diabetic care support, and implemented the solution in 1,146 retail pharmacies in 94 cities in China. Dia- WITHDRAWN betic patients who were pharmacy members can receive in-store services of: 1) BG testing with the results synchronized to the cloud, 2) App-assisted personalized coaching, 3) App-assisted personalized meal plans, and 4) pharmacy consultation. This report profiles a diabetic population of 85,790 during Jun 6th 2015 and Nov 4th 2016. Results: The study population had an average age of 59.7 ± 11.8, BMI of 23.7 ± 3.4, and male of 44.3%. A total of 204,887 BG tests were performed, and 28,492 people had at least 2 measurements. Both fasting blood glucose (FBG) and random blood glucoses (RBG) were improved (Figure). Plotting the baseline FBG readings against the category of each city’s economic status found a strong negative correlation r = -0.92, P = 0.029, with the lowest baseline FBG in the economically most advanced cities. Yet the same analysis for the glycemic control effect indicated no correlation r = -0.14, P = 0.817. Conclusion: Regional economic status affects the population BG level and pharmacists are effective in community diabetic control regardless the geographic difference.

ADA-Supported Research For author disclosure information, see page A751.

A613 CLINICAL THERAPEUTICS/NEW TECHNOLOGY—GLUCOSECATEGORY MONITORING AND SENSING

2352‑PUB 2354‑PUB A Chinese Study to Build a HbA1c Prediction Model with 12-Week Mid-infrared Quantum Cascade Laser Spectroscopy for Noninva‑ Self-Monitoring Blood Glucose Values sive, In Vivo Glucose Sensing LING WANG AN, YOU JIE ZHANG, JU MING LU, ZHAO HENG HU, YAU JIUNN ALEXANDRA WERTH, SABBIR LIAKAT, CLAIRE GMACHL, Princeton, NJ LEE, LI NONG JI, A1C PREDICTION STUDY GROUP, Beijing, China, Shangluo, China, Quantum cascade (QC) lasers, invented in 1994, have drastically expanded Pingtung, Taiwan the opportunities for mid-infrared spectroscopy. The mid-infrared region, Objective: The HbA1c value is widely used to judge the diabetes control. often called the fingerprint region, has strong molecular absorption features Day-to-day diabetes management is guided by self-monitoring blood glucose for a multitude of biomarkers. QC laser spectroscopy can provide sensitive (SMBG) data. To predict HbA1c with daily SMBG data may encourage people and selective monitoring of these biomarkers. Recently, we have imple- with type 2 diabetes mellitus (T2DM) to pursue good glycemic control. This mented a noninvasive, mobile glucose sensor based on this technology. Our retrospective study aimed to define a mathematic model to predict HbA1c system has three main components: a QC laser, an integrating sphere, and a value with SMBG values obtained from daily lives. thermal-electrically cooled mercury cadmium telluride (MCT) detector. The Methods: A total of 195 subjects with T2DM with 245 HbA1c and 21,375 QC laser is swept from 8-10µm; this wavelength range contains unique spec- SMBG data were analyzed. HbA1c data were collected after the 12 weeks tral absorption features of glucose, particularly the C-O stretching mode at SMBG. One HbA1c value was correspondent with a mean of 87 SMBG values. 9.5µm. The light penetrates into the dermis layer of the skin where it is Results: By linear regression analysis, there were significant correlation absorbed by the glucose molecules in the interstitial fluid. The light is then observed between the HbA1c level and past three 28-days average SMBG backscattered off of the collagen fibers and other scatters then collected values (A1C=2.986+0.119×SMBG3+0.198×SMBG2+0.167×SMBG1, R2=0.432, using the integrating sphere and MCT detector. Backscattered spectra are p=0.000), allowing prediction of HbA1c with changing SMBG. The linear collected every 5 minutes from a single subject after eating a meal. We ana- regression equation between the HbA1c and the 84-days SMBG values also lyze the series of collected spectra using principal component (PC) analysis showed good correlation (y=3.119+0.466×SMBG, R2=0.432, p=0.000). When to determine the wavelengths that correspond to highest variance. We con- compared the estimated HbA1c and observed HbA1c values, 70% of the sam- sistently see a strong correlation with the predicted PC of the spectra and ples had variance within 10% and 88% of the samples had variance within the known glucose absorption spectrum. 15%. Similar results were found when compared the estimated HbA1c calcu- Figure. lated by equation provide by Nathan DM (and A1c-Derived Average Glucose Study Group. Diabetes Care,2008,31:1473-1478) and the observed HbA1c Therapeutics values. The samples with higher variance had significantly fewer SMBG Clinical Diabetes/ frequency (p=0.011). PUBLISHED ONLY Conclusions: This HbA1c prediction model could be experimentally used for subjects with T2DM in daily diabetes management. However, the variance between the estimated HbA1c and real HbA1c value may be big when the SMBG frequency is low.

2353‑PUB The Lies HbA1c Tells GABRIEL UWAIFO, JENEE NGUYEN, New Orleans, LA, Slidell, LA HbA1c is used for diagnosis and monitoring of diabetes (DM) but there are limitations to its accuracy. We present three cases that highlight the need for caution in HbA1c use in clinical care. Case 1 is a 59 yr old African American man with > 12 yr history of type 2 DM and sickle cell disease. Past records showed normal HbA1c (4.5-5) and near normal glycoHemoglobin (5.4-7.4) Supported By: Schmidt Family Foundation; MIRTHE (EEC-0540832) despite mean blood glucose (BG) > 180mg/dl. At initial visit with us attempts to obtain HbA1c were impossible due to finding of a hemoglobin variant that 2355‑PUB influenced both ion exchange and boronate affinity HPLC. Instead, his gly- Continuous Glucose Monitoring: A Training and Treatment Program cemic profile is tracked using BG, Fructosamine and glycomark. Hemoglobin for All Age Groups electrophoresis (HBE) showed HBSC disease. Case 2 is a 72 yr old Caucasian ULRIKE THURM, BERNHARD GEHR, MARTIN HOLDER, BERND KULZER, KARIN (C) lady referred because of discrepancies between HbA1c and BG. In the LANGE, ANDREAS LIEBL, CLAUDIA SAHM, SIMONE VON SENGBUSCH, THOR- last year HbA1cs were 4.0-4.8 despite BG values in the 106-185mg/dl range. STEN SIEGMUND, RALPH ZIEGLER, GUIDO FRECKMANN, SANDRA SCHLÜTER, OGTT showed impaired fasting glucose. Prior HbA1cs were done using immu- LUTZ HEINEMANN, Ulm, Germany noassay. At our review repeat HbA1c was sent to Mayo labs. This showed Patients using a system for continuous glucose monitoring (CGM) without an interfering substance affecting ion-exchange HPLC. Her sample was special training often do not achieve their intended improvement of meta- measured using boronate affinity HPLC and HbA1c was 6.0. HBE revealed bolic control; training means translation of the information provided by the HB J- Baltimore. Case 3 is a 46 yr old C lady with morbid obesity. Her HbA1cs CGM system in appropriate therapeutic action and not so much handling were in the 5.6-5.9 range and she enrolled in a lifestyle modification plan of the system. Reimbursement for CGM is provided in Germany only if the with metformin 500mg BID. After ~ 1 yr she developed intermittent bilat- patients are trained adequately. A team of about 20 diabetes nurses, diabe- eral leg paraesthesiae due to small fiber peripheral sensory neuropathy. She tologists and psychologists from the German Diabetes Technology working also had sudden onset visual blurring. Ophthalmology review showed right group (AGDT) and the Pediatric Diabetology working group (AGPD) of the retinal hemorrhage with partial retinal detachment and bilateral background German Diabetes Association (DDG) has developed a structured and inde- retinopathy consistent with diabetic retinopathy. OGTT done after holding pendent CGM training program called “SPECTRUM” (Structured patient edu- metformin showed DM. She was commenced on liraglutide and metformin cation and treatment program for self-reliant continuous glucose monitor- dose increased to 1000mg BID. HBE revealed a variant of HB Barts. HbA1c ing), to ensure optimal usage of CGM. SPECTRUM consists of specific sets of should be interpreted with accompanying BG measurements. When a dis- slides for adults (introductory session + 6 training modules), for parents with crepancy is found other indices like fructosamine and glycomark can be use- younger children and adolescents (introductory session + 5 training modules ful. Repeat HbA1cs with other methods may help identify artefactually high for each version), a structured curriculum and a SPECTRUM folder. The pro- or low HbA1cs and so guide appropriate clinical care. gram does not favor any given CGM system. Implementation of SPECTRUM has started in 2016; 245 diabetologists and CDEs were trained in using the adult version and 112 in the paediatric version. We have 364 health profes- sionals on the waiting list for 2017. This reflects the high needs for such a training program. Anecdotical patient reports highlight that just providing the technology is not sufficient, patients (and their treating diabetes team) need a good training to make optimal usage of this modern but expensive diagnostic option. Currently we prepare a study (CGM TRAIN) to evaluate systematically how much patients benefit from participation in such a training program.

For author disclosure information, see page A751. ADA-Supported Research

A614 CLINICAL THERAPEUTICS/NEWCATEGORY TECHNOLOGY—INSULINS

2356‑PUB CLINICAL THERAPEUTICS/NEW TECHNOLOGY— Impact of Continuous Blood Glucose Monitoring on Patient Empow‑ INSULINS erment, Distress, and Glycemia in a High Risk Population ARTI A. THANGUDU, DIANE BATTAGLIA, LINDA M. SIMINERIO, SANDRA I. 2358‑PUB SOBEL, Pittsburgh, PA Profiles of Patients with Type 2 Diabetes Needing 1 to 3 Injections of Background: An aim of diabetes management is to help empower patients Insulin Lispro Mix25 in PARADIGM Study with diabetes mellitus (DM) to interpret blood glucose monitoring (BGM) KEITH BOWERING, DACHUANG CAO, JOÃO FELICIO, LINONG JI, HENRY SCHMITT, results to improve glycemia and reduce distress. Team-based care, educa- INDRANIL BHATTACHARYA, ABDUL JABBAR, Edmonton, AB, Canada, Indianapolis, tion and continuous glucose monitoring (CGM) have been shown to improve IN, Belém, Brazil, Beijing, China, Brussels, Belgium, Gurgaon, India, Dubai, United Arab glycemia, however, are not always available in high risk underserved com- Emirates munities. Background: Few countries show interest in using simplified insulin mix- Objective: Implement a program in a population with limited access to ture regimen with progressive intensification to decrease clinical inertia in technology, team care and education to improve patient empowerment and patients on oral antidiabetic drugs (OADs). glycemia while reducing distress. Methods: PARADIGM was a 48-week study in patients inadequately con- Methods: Fourteen individuals with T1DM or T2DM on insulin therapy trolled on OADs who were randomized to 1-3 daily injections (inj) of insulin with a HbA1c >8% and/or hypoglycemia unawareness, were invited to meet lispro Mix25 or insulin glargine alone or with 1-3 daily inj of insulin lispro. with an endocrinologist who reviewed data and a diabetes educator (DE) for In this post-hoc analysis, glycemic profiles of lispro Mix25-treated patients education on use and interpretation of CGM results. Devices were worn for were analyzed by number of inj (N=38, 72, 63) with study sequence (din- seven days. The Empowerment Scale and Problem Areas in Diabetes (PAID) ner, breakfast, lunch) and by ethnicity (Caucasians=36; Asians=77; Hispan- surveys were administered to test empowerment and diabetes-distress. ics=60). HbA1c and surveys were evaluated at baseline and three months. Results: At baseline, pre- and postprandial glycemic levels (Figure) and Results: Patient empowerment and HbA1c improved. While diabetes dis- A1C (1 inj=8.8, 2 inj=8.9, 3 inj=9.1%) were higher with more inj. At week 48, tress worsened. glycemic profiles and A1C (7.3, 7.2, 7.3%) were similar for 3 insulin regimens. Conclusions: A comprehensive program that includes CGM, team care Total daily insulin dose increased with number of inj (23.0, 51.4, 78.6 u/d), but and education afforded improvements in glycemia to an underserved high hypoglycemia rate decreased (33.4, 21.3, 16.3 events/patient/year). Similar risk population with limited access to current technology and quality care.

trends were observed across ethnicities; however, with different propor- Therapeutics While empowerment scores and glycemia improved, patients experienced

tions of patients on 1-3 insulin doses and hypoglycemia rates. Clinical Diabetes/ distress. The exposure to real-time BG data and heightened diabetes aware- Conclusion: Progressive increase in number of inj of lispro Mix25 is an PUBLISHED ONLY ness needs to be considered and further examined in a larger sample of high alternative simple insulin regimen allowing patients of different ethnicities risk patients. to reach good glycemic control with minimal number of inj and no increase Supported By: The Beckwith Institute in hypoglycemia rate. 2357‑PUB Is Fructosamine a Better Biomarker for Glycemic Control in Patients with Diabetes Mellitus? JANICE L. GILDEN, AMENA IQBAL, MAHWASH SIDDIQUI, SRIKAR RAPAKA, DEGAULLE DAI, ALI LADHA, BOBY G. THECKEDATH, ALVIA MOID, North Chicago, IL Control of blood glucose has been shown to decrease diabetic complications. Although, many studies have used glycosylated hemoglobin (HbA1c), an integrated value for 2-3 months of blood glucose levels, as a measure of glycemic control, several conditions, such as anemia and hemoglobinopathies, and chronic renal failure (CKD), can result in falsely higher values. Certain ethnic groups have also been observed to have higher levels of glycosylation and are not thought to have diabetes mellitus (DM). In addition, this measure does not reflect glucose variability, which may be a more important indicator, nor does it reflect recent changes in glycemic control, leading many providers to increase medications to achieve a lower value too rapidly in order to meet quality indicators, at the expense of creat- Supported By: Eli Lilly and Company ing hypoglycemia. Unlike HbA1c, which measures the glycosylation of the hemoglobin A1 fraction with a life-span of 120 days, fructosamine measures 2359‑PUB glycosylated plasma proteins, and thus reflects glycemic control over the Predictive Factors for Achieving Glycemic Control in People with past 2-3 wks. In addition, fructosamine is not subject to variation by these Type 2 Diabetes (T2D) in Real Clinical Practice same conditions. Therefore, we compared fructosamine to HbA1c in 140 TIRTHANKAR CHAUDHURY, ABDULQAWI AL MANSARI, YOUSSEF OBEID, DM patients [(age range=24-85 yrs.) (9 type 1: 131 type 2) (17 female: 123 NAJAMUL ISLAM, MOHAMMED FARIDUDDIN, AHMED HASSOUN, KHIER DJA- male) (2% anemia/hemoglobinopathies: 31% CKD: 45% recent changes in BALLAH, MOJTABA MALEK, Kolkata, India, Jeddah, Saudi Arabia, Beirut, Lebanon, medical management)]. Results demonstrated that there was a significant Karachi, Pakistan, Dhaka, Bangladesh, Dubai, United Arab Emirates, Paris, France, discordance between the expected HbA1c based upon the fructosamine Teheran, Islamic Republic of Iran [(0.017 X fructosamine + 1.61)], as well as a significant difference between Current ADA/EASD position statement recommend to individualize both the expected fructosamine based upon the HbA1c [(HbA1c-1.16) X 58.82] for treatment targets and treatment strategies in inadequately controlled pts the entire cohort (p<0.001). In addition, there was a negative correlation by oral glucose-lowering drugs (OGLDs) to decrease the burden of diabetes- between the stage of CKD and the difference between actual and predicted related complications. We conducted a 12 months real-world practice study HbA1c (p<0.001). from October 2012 to January 2015 in 10 developing countries. The main In conclusion, fructosamine may be used as another biomarker for assess- objective of the study was to assess predictive factors for achieving the ing the current level of glycemic control in patients with DM, especially glycemic HbA1c at 6 months targeted by the treating physician in adults with when recent changes in management have been made, as well as when T2D requiring insulin initiation, titration and/or intensification. There were other common co-morbid conditions exist. 2,704 included eligible pts (mean (SD) age: 54.6 (10.6) years, BMI: 28.7 (5.4) Supported By: Captain James A. Lovell Federal Health Care Center kg/m2; Caucasian: 46.1%, T2D duration: 10.1 (6.7) years) with poor glycemic control (mean (SD) HbA1c: 9.7 (1.8)%, FBG: 196.8 (60.4) mg/dL). At baseline, 63.3% were treated with OGLDs alone, 27.9% with OGLD + insulin and 8.8% with insulin alone (basal [B]: 38.8%, premix: 35.6%, basal + prandial [BP]: 17.1%; N=931). From baseline to the end of study, the proportion of pts receiving OGLDs + insulin, B, and BP increased (85.4%, 71.2%; and 25.1% respectively) as the mean daily doses of B (16.5 to 21.9 U) and BP (56.1 to 61.9 U). At 6 months, advanced age, Caucasian ethnicity, shorter duration of T2D

ADA-Supported Research For author disclosure information, see page A751.

A615 CLINICAL THERAPEUTICS/NEWCATEGORY TECHNOLOGY—INSULINS

(>10 vs. 1 year), lower HbA1c at baseline (≥ 8.5% vs. <7%) (p≤0.0001) and no 2362‑PUB intake of OGLD (none vs. 2, p=0.02) were predictive factors for achieving gly- The Impact of Baseline BMI and HbA1c on Glycemic Control after cemic goal as targeted by the treating physician. Absence of high blood pres- Treatment with Mealtime Fast-Acting Insulin Aspart in People with sure (p=0.0479) and use of basal insulin therapy (p<0.0001) were additional Type 1 Diabetes significant predictors identified at 12 months. Absolute changes in the mean DAVID RUSSELL-JONES, SIMON R. HELLER, VINCENT C. WOO, VINAY BABU, HbA1c of -1.7% and -2% were observed from baseline to 6 and 12 months CLAUS DETHLEFSEN, CHANTAL MATHIEU, Guildford, United Kingdom, Sheffield, respectively. Along with some well-known predictive factors (shorter T2D United Kingdom, Winnipeg, MB, Canada, Søborg, Denmark, Leuven, Belgium duration, lower baseline HbA1c, no hypertension), this real world study in Onset 1 was a 26-week, randomized trial evaluating the efficacy and developing countries suggested that insulin regimen treatment initiation safety of fast-acting insulin aspart (faster aspart) in adults with type 1 diabe- and/or intensification allowed pts to achieve a better glycemic control. tes (T1D). Patients were randomized to double-blind mealtime faster aspart Supported By: Sanofi (n=381), insulin aspart (IAsp; n=380) or open-label post-meal faster aspart (n=382); each with insulin detemir. This post-hoc analysis investigated the 2360‑PUB impact of baseline body mass index (BMI; subgroups: <25, 25-30, ≥30 kg/ 2 Insulin Pharmacodynamics during Follicular and Luteal Phases of m ) and HbA1c (subgroups: ≤7.5%, 7.5-8.0%, ≥8.0%) on glycemic control with Menstrual Cycle mealtime faster aspart and IAsp. EDA CENGIZ, WILLIAM TAMBORLANE, NEHA PATEL, JENNIFER SHERR, In the overall population, change in HbA1c after 26 weeks was non-inferior MICHELLE VANNAME, AMY STEFFEN, JENNIFER FINNEGAN, EILEEN TICHY, LORI (0.4% limit) for mealtime faster aspart vs. IAsp, with an estimated treatment CARRIA, ANDREA URBAN, SIRMEN KIZILCAN, SONALI SAXENA, NITIN SUKU- difference (ETD [95% CI]) of −0.15% (−0.23;-0.07). ETD for change in HbA1c MAR, STUART WEINZIMER, New Haven, CT, Izmir, Turkey was similar across analyzed BMI and HbA1c subgroups (Table). No major dif- Impact of menstrual cycle on insulin action in females with type 1 dia- ferences between treatments were observed for severe or blood glucose betes. Insulin action during different phases of the menstrual cycle (MC) (BG)-confirmed hypoglycemia across subgroups (Table). Total daily insulin has not been well investigated and is associated with unpredictable blood dose was similar in patients across all baseline HbA1c groups and the BMI glucose fluctuations each month in females with type 1 diabetes (T1D). To <25 or 25-30 kg/m2 groups, but was significantly lower with mealtime faster assess how the phase of the MC affects insulin action, we characterized the aspart compared with IAsp in subjects with baseline BMI >30 kg/m2 (Table). pharmacodynamic properties of rapid-acting insulin analogs using the eug- In conclusion, treatment difference between faster aspart and IAsp for lycemic clamp technique after a single, subcutaneous dose of insulin aspart glycemic control in people with T1D was not affected by baseline BMI or Therapeutics (0.2u/kg) in a randomized order, cross-over study conducted during both the baseline HbA1c. Clinical Diabetes/

PUBLISHED ONLY luteal phase (LP) and follicular phase (FP). Sixteen subjects have been enrolled, and to date, five (age 23±8 yrs, HbA1c 7.0±1%, duration of T1D 14±8 yrs, insulin dose 0.8±0.1 u/kg/day) have completed both insulin action studies. As demonstrated in the Table, the average glucose infusion rate (GIRmean), maximum glucose infusion rate (GIRmax), and the area under the curve for GIR (AUCGIR 0-300min) are significantly decreased during the LP as compared to the FP. The time to reach maximum GIR (TGIRmax) was not significantly different. These preliminary results describe significant differences in insulin pharmacodynamics during different phases of the MC, with almost 50% reduction in glucodynamic action during LP, underscoring the importance of devising MC adjusted insulin treatment for better management of T1D in females. Table. Pharmacodynamic Measures Follicular Phase Luteal Phase p (mean±SD), (n=5)

GIRmean (mg/kg/min) 3.7±0.9 2.2±0.8 0.008

GIRmax (mg/kg/min) 6.8±2.0 4.5±1.2 0.001

T GIRmax (min) 84±46 110±51 0.5

AUC GIR 0-300min (mg/kg) 1128±282 575±74 0.007 Supported By: Novo Nordisk A/S Supported By: The Leona M. and Harry B. Helmsley Charitable Trust 2363‑PUB 2361‑PUB The Mechanics of Subcutaneous Insulin Absorption WITHDRAWN SPENCER T. FRANK, LING HINSHAW, RITA BASU, ANDREW J. SZERI, ANANDA BASU, Berkeley, CA, Rochester, MN A key component of a successful artificial pancreas controller is its ability to predict the rate of insulin absorption. Particularly, if subcutaneously injected insulin reaches the bloodstream significantly faster or slower than nominal rates, hypoglycemia or hyperglycemia may result. For this reason we study the factors that impact the rate of insulin absorption from the subcutaneous tissue by relating tissue perfusion (i.e., tissue blood flow) to insulin absorption rates. We validate this relationship on two cohorts of T1D subjects: resting subjects (n=20, age = 42 ± 14 years, hemoglobin A1c = 7.2 ± 0.6%) and subjects exercising at 50% V02 max for 75 minutes (n=14, age = 45 ± 13 years, hemoglobin A1c = 7.6 ± 0.7%). Each cohort underwent a mixed meal tolerance test with a subcutaneous bolus of insulin pumped at mealtime with plasma insulin concentrations measured against time. Utilizing average tissue perfusion rates and average capillary permeability found in literature, our early results indicate that insulin concentration predictions for an average resting patient and an average exercising patient are close to experiments. Additionally, we show that tissue blood perfusion rates during exercise account for the acceler- ated rate of insulin absorption-an important phenomenon to capture because it helps to ensure safe glucose ranges during and after exercise. Supported By: National Institutes of Health; National Science Foundation

For author disclosure information, see page A751. ADA-Supported Research

A616 CLINICAL THERAPEUTICS/NEWCATEGORY TECHNOLOGY—INSULINS

2365‑PUB Efficacy and Safety of Insulin Degludec (IDeg) vs. Insulin Glargine U100 (IGlar) in Hispanic Patients with Type 2 Diabetes (T2D) LOUIS B. CHAYKIN, ANUJ BHARGAVA, CAROL H. WYSHAM, TRINE ABRAHA- MSEN, MICHIEL VAN LEEUWEN, RAYMOND DE LA ROSA, Bradenton, FL, Des Moines, IA, Spokane, WA, Søborg, Denmark, Paducah, KY The risk of developing T2D is higher in Hispanic populations compared with non-Hispanic Caucasians. The safety and efficacy of IDeg once daily (OD) vs. IGlar OD was assessed post-hoc in a subgroup of Hispanic patients with T2D in a 64-week, double-blind, treat-to-target, crossover trial (SWITCH 2; NCT02030600). Of the patients in SWITCH 2 (n=721), 36% were Hispanic (IDeg/IGlar n=140; IGlar/IDeg n=122). Baseline characteristics including A1c were comparable between treatments. A1c reductions with IDeg were similar to IGlar (ETD 0.03% [-0.21; 0.27]95% CI and 0.15% [-0.09; 0.39]95% CI after period 1 and 2, respectively). The rate of overall symptomatic (severe [requiring third-party assistance and external adjudication] or blood glucose [<56 mg/dL] confirmed) hypoglycemia in the maintenance period (after a 16-week titration period) was numerically lower with IDeg than IGlar. The rate of nocturnal symptomatic (00:01-05:59, both inclusive) hypoglycemia in the maintenance period was significantly lower with IDeg vs. IGlar; p<0.05 (Figure). Results were similar in the full treatment period. Adverse events (AE) and serious AEs were comparable between treatments. In conclusion, results in a Hispanic subpopulation are consistent with the overall trial results; IDeg and IGlar lead to good glycemic control with a lower risk of hypoglycemia with IDeg vs. IGlar in Hispanic patients with T2D. Therapeutics Clinical Diabetes/ PUBLISHED ONLY

2364‑PUB Anti-insulin Antibodies and Immune Adverse Events with SAR342434 Insulin Lispro Compared with Humalog Insulin Lispro in People with Diabetes PHILIP D. HOME, KARL-MICHAEL DERWAHL, MONIKA ZIEMEN, KARIN WER- NICKE-PANTEN, SUZANNE PIERRE, YVONNE KIRCHHEIN, SATISH K. GARG, Newcastle upon Tyne, United Kingdom, Berlin, Germany, Frankfurt, Germany, Paris, France, Aurora, CO SAR342434 (insulin lispro, Sanofi) is a follow-on (biosimilar) of insulin lispro (Humalog®, Lilly). Both phase 3 studies in people with type 1 (T1DM; 12-month study, SORELLA 1) and type 2 diabetes (T2DM; 6-month study, SORELLA 2) were randomized (1:1), open label, parallel-group studies comparing the efficacy and safety of these two insulins in combination with insulin glargine (U100; Lantus®). Baseline antibody positivity was similar for both insulins, but higher in T1DM than in T2DM. In both studies, the percentage of people newly positive for AIAs in the two treatment groups, or having a ≥4-fold increase in AIA titer, was not found to differ (Table). No relationship was observed between highest individual titers and HbA1c change, insulin dose, hypoglycemia, hypersensitivity or injection site reactions. Formal adjudications of increase in HbA1c >1.0% or unexplained increase in insulin dose found no association with AIAs. Hypersensitivity events and events Supported By: Novo Nordisk A/S adjudicated as allergic reactions were few and did not differ between the two insulins (Table); none of them was suspected to be AIA mediated. Over- 2366‑PUB all efficacy/safety results in the studies did not differ (presented elsewhere). Pooled Analysis of Seven Randomized Controlled Trials with Insulin We conclude the follow-on insulin lispro SAR342434 had a similar immuno- Glargine 100 U/mL Stratified by Type of Meal-Time Insulin in Adults genicity profile in people with T1DM and T2DM to originator insulin lispro. with T1DM GEREMIA B. BOLLI, DAVID R. OWENS, GREG FULCHER, BRIAN M. FRIER, MEI ZHANG, WOLFGANG LANDGRAF, DEMET OZKAYA, PHILIP D. HOME, Perugia, Italy, Cardiff, United Kingdom, Sydney, Australia, Edinburgh, United Kingdom, King of Prussia, PA, Frankfurt, Germany, Paris, France, Newcastle upon Tyne, United Kingdom In order to define more precisely the efficacy and safety outcomes of insulin glargine 100U/ml (Gla-100) in people with T1DM, standardized patient- level data were pooled from seven RCTs (26-30 weeks duration) when using QD Gla-100 together with either regular human insulin (RHI) or rapid-acting insulin analogs (RAI). HbA1c, FPG, body weight, insulin dose and hypoglycemia were assessed descriptively through to study endpoint. Overall, 1,282 participants (49% male) on Gla-100 and prandial insulin (RHI n=694, RAI n=580, unknown n=8) were included. They were (mean ± SD) aged 41 ± 13 years, diabetes duration 16 ± 11 years, and BMI 24.5 ± 3.8 kg/m². Mean baseline and endpoint Gla-100 doses were similar (0.31 ± 0.15 and 0.29 ± 0.13 U/kg), while total insulin dose was 0.70 ± 0.23 U/kg at baseline and did Supported By: Sanofi not change. Clinical outcomes are shown in the Table. HbA1c reduction was small with both RHI and RAI, with a low percentage (14%) of people reaching

ADA-Supported Research For author disclosure information, see page A751.

A617 CLINICAL THERAPEUTICS/NEWCATEGORY TECHNOLOGY—INSULINS

target of <7.0%. Hypoglycemia affected a high proportion of people (overall, and SS on various days, n=2459). In our first analysis, we identified factors 89%; nocturnal 61%), but hypoglycemia requiring assistance only 11%. This associated with choice of insulin therapy. Now we evaluate glucose control pooled analysis in adults with T1DM confirms that, in the presence of a high and patient outcomes with a propensity score analysis adjusting for those incidence and rate of hypoglycemia (overall, nocturnal), only a minority of factors. A patient day was defined by point of care glucose (POC) in mg/dL people reached glycemic targets, likely due to insufficient optimization of as hypoglycemic (<70), hyperglycemic (mean >180) or euglycemic (none <70 basal insulin and independent of type of meal-time insulin. and mean ≤180). Patient outcomes were mortality, length of stay (LOS) and readmission. Negative binomial regression analysis was used to determine unique factors associated with count of hyper-, hypo- and euglycemic days; linear regression was used to model log of LOS; logistic regression modeled 30 and 60 day readmission rates. Average POC was 30 mg/dL less for SS and 24 mg/dL less for Mixed than BB. SS had less hypo- and hyperglycemic days (P<0.001), while Mixed had more hypoglycemic days (P<0.001) than BB. Mortality rates were: 0.0% BB; 0.07% SS (n=1); 0.45% Mixed (n=11). BB had similar average LOS as SS but 1.7 days (32%) lower than Mixed (P<0.0001). Readmission rates did not differ across groups. BB was used in more complex patients, and had worse glycemic control than SS, but with equivalent patient outcomes. BB had less hypoglycemia but not hyperglycemia, and significantly lower LOS than Mixed, suggesting that glycemic control alone may not fully explain the improved outcomes with BB. More investigation is needed to understand the effects of BB, particularly in mixed insulin regimens.

2369‑PUB Body Mass Index and Age Are Independent Determinants for Basal Insulin Requirement in Patients with Type 1 Diabetes Supported By: Sanofi AKIO KURODA, MUNEHIDE MATSUHISA, Tokushima, Japan Background and Aim: The determinant for basal insulin requirement in the Therapeutics 2367‑PUB daily life has not been well documented in patients with type 1 diabetes. We Clinical Diabetes/

PUBLISHED ONLY Observational Registry of Basal Insulin Treatment (ORBIT) in investigated the basal insulin dose requirement and associated factors in Patients with Type 2 Diabetes in China: Safety and Hypoglycemia patients with type 1 diabetes. Predictors Methods: Fifty-six inpatients with type 1 diabetes treated with sub- TINGTING ZHANG, LINONG JI, YAN GAO, PUHONG ZHANG, DONGSHAN ZHU, cutaneous insulin injections were enrolled. All patients were hospital- HENG ZHANG, XIAN LI, JIACHAO JI, XIAOHUI GUO, Beijing, China ized because of blood glucose control and education from 2011 to 2016 in Background: Basal insulin (BI) has been widely used in clinical practice Tokushima University Hospital and were obtained the written informed for diabetes therapy. The Observational Registry of BI Treatment (ORBIT) consent for this study. They were treated with foods under the control of study was designed to evaluate the safety of BI after initiation in real-world dietitian. Basal insulin dose were adjusted to maintain the blood glucose at practice settings in China. the same values before sleep and before breakfast. Clinical characteristics Methods: Patients with type 2 diabetes mellitus (T2DM) inadequately and basal insulin requirement were investigated. controlled with oral hypoglycemic agents (OHAs) from 8 geographic regions Results: Twenty-two male and 34 female patients were enrolled (Age, and 2 hospital tiers in China initiated BI treatment. Data for body weight and 53.0±16.4 years; BMI, 23.1±4.2; HbA1c 8.9±1.9%). Total daily dose of insulin hypoglycemic episodes were collected at baseline, 3 and 6 months. Serious (TDD) was 36.1±16.9 units (0.62±0.22 units/kg). Total daily basal insulin dose adverse events (SAEs) were collected at 3 and 6 months. was 9.2±6.1 (25% of TDD) units. Plasma C-peptide was 0.29±0.46 ng/mL. Results: BI use did not induce significant weight gain; long-acting basal Total insulin requirement in insulin-deficient (C-peptide<0.2ng/mL) patients insulin analogs, especially insulin detemir, caused less weight gain than that were 32 among 56 patients and total basal insulin dose was 27.2±0.11% of of NPH insulin. At the end of this study, general hypoglycemia was reported TDD. The ratio of total basal dose to TDD was significantly correlated with in 7.7% and severe hypoglycemia was 0.3%. General hypoglycemia had a BMI (r=0.361, p<0.01), plasma C-peptide (r=-0.334, p<0.05), age (r=-0.363, slightly increasing trend (from 1.62 up to 1.79 times/patient/year) and severe p<0.05), and HbA1c (r=0.277, p<0.05). BMI (p<0.01) and age (p<0.05) were hypoglycemic event had a slightly decreasing trend (from 0.05 down to 0.03 the significant factors for the determinant of % basal insulin dose according times/patient/year). Probable symptomatic hypoglycemia events had the to the multiple regression analysis. highest proportion of total hypoglycemia events. The increase of hypoglyce- Conclusion: Low basal insulin requirement was associated with good gly- mia was mainly seen in patients taking NPH insulin and insulin detemir. Age, cemic control, and was determined by low BMI and age. inpatient or outpatient status, body mass index, HbA1c at baseline and end of study, T2DM duration, microvascular complications, BI type, combination 2370‑PUB with insulin secretagogues, SMBG times, and insulin dosage all predicted Baseline Red Blood Cell Distribution Widths Indicate Long-Term hypoglycemia. In total, 3.5% of patients have at least one SAE during the Glycemic Remission in Patients with Type 2 Diabetes study period. Most SAEs of patients (86.8%) were deemed unrelated to LIJUAN XU, LIANGJIAO WANG, XINWEI HUANG, YANBING LI, Guangzhou, China OHAs or insulin treatment. Aims: Short-term continuous subcutaneous insulin infusion (CSII) has been Conclusions: In the ORBIT study, BI use, especially that of insulin detemir shown to be effective for inducing long-term drug-free euglycemic remission and insulin glargine, was associated with minimal weight gain and good in nearly half patients with newly diagnosed type 2 diabetes; yet, predictors safety outcomes for hypoglycemia in patients with T2DM in real-world set- of remission are still unsettled. We explored whether red blood cell distribu- tings of China. tion widths (RDWs), a routinely checked parameter of complete blood cell Supported By: Sanofi China counts, was an indicator of long-term euglycemia remission. Methods: We analyzed the original data from patients enrolled in three 2368‑PUB randomized control trials from 2002 to 2014 with similar inclusion and Basal-Bolus vs. Other Insulin Strategies in a “Real World” Hospital exclusion criteria. CSII was administered to drug-naïve patients with newly Setting: Association with Glucose Control and Patient Outcomes diagnosed type 2 diabetes. The insulin dosage was titrated to achieve and ARCHANA R. SADHU, AISHA VADHARIYA, BHARGAVI PATHAM, HARLAN maintain euglycemia for 2 weeks. SPARROW, MICHAEL L. JOHNSON, Houston, TX Results: A total of 185 patients were involved. Ninety-eight patients Guidelines on inpatient hyperglycemia in noncritically ill patients recom- (52.97%) who achieved and maintained euglycemia for at least 12 months mend using basal bolus insulin (BB) therapy instead of sliding scale (SS) were classified as the remission group, and the other 87 patients as the alone. However, limited outcomes data exist from large scale “real world” non-remission group. Patients in remission group had a relatively lower settings. We retrospectively reviewed 4,670 admissions between Jan 2013 value for RDWs (38.82±2.76 vs. 39.89±2.78 fL, p=0.017) compared with and Sep 2015. Patients ≥18 years who received subcutaneous insulin ≥ 75% those in non-remission group at baseline. A graded decrease of remission of the hospital stay but not any IV insulin were included. Insulin therapy was rate (67.50%, 55.00%, 53.66% and 30.77% for Quartile 1 to Quartile 4 grouped into 3 types: BB only (n=842), SS only (n=1428), or Mixed (both BB respectively, P<0.05) was observed with the increasing of RDWs. The risk

For author disclosure information, see page A751. ADA-Supported Research

A618 CLINICAL THERAPEUTICS/NEWCATEGORY TECHNOLOGY—INSULINS of hyperglycemic relapse was significantly increased for those in the high- an increase of 5-8 U (27.2% vs. 20.1%; p=0.03) and > 8 U (26.2% vs. 20.1%; est quartile compared with the lowest (hazard ratio=2.68; 95% CI, 1.38 to p=0.02) compared to no titration. Confirmed, symptomatic hypoglycemia 5.22). In addition, Patients who achieved euglycemia within 7 days (hazard was documented in 1.5%, 2.4%, 1.2%, and 2.2% of pts in ascending order of ratio=0.60, p=0.046) or a better fasting glucose (6.12±1.38 vs. 6.67±1.29 titration with no difference between groups. The majority of pts in clinical mmol/L, p<0.001) after therapy had a preferable remission rate. practice were titrated slow and treatment was not intensified early enough Conclusions: Patients with lower baseline RDWs are more likely to obtain and stopped as soon as a fasting blood glucose of about 130 mg/dl was a one-year euglycemia remission after short-term CSII. A faster normaliza- achieved. tion of glucose during treatment and a lower fasting glucose after therapy Supported By: Sanofi Aventis Deutschland GmbH are beneficial for improving the long-term glucose control. 2373‑PUB 2371‑PUB Insulin Requirement in Well-Controlled Type 1 Diabetes Patients The Impact of Age on Real-World Titration of Insulin Glargine 100 with Continuous Subcutaneous Insulin Infusion or Multiple Daily U/mL in Patients with Type 2 Diabetes Poorly Controlled on Oral Injection Therapy Antidiabetic Drugs PING LING, JINHUA YAN, QINGWEI XIE, LILING QIU, SIHUI LUO, XUEYING ZHENG, ANDREAS FRITSCHE, HELMUT ANDERTEN, ANJA BORCK, KATRIN PEGELOW, HEYING AI, DAIZHI YANG, YAN WU, JIANCAI LIN, FUNENG WANG, JIANNENG MARTIN PFOHL, STEFAN PSCHERER, JOCHEN SEUFERT, Tübingen, Germany, WU, SHAOQING LI, BIN YAO, JIANPING WENG, Guangzhou, China Hildesheim, Germany, Berlin, Germany, Duisburg, Germany, Weimar, Germany, Objective: Continuous subcutaneous insulin infusion (CSII) or multiple Freiburg, Germany daily injection (MDI) are the recommended therapies for patients with type 1 Adding insulin glargine 100 U/ml (Gla-100) improves glycemic control in diabetes (T1D). The aim of this study was to investigate the insulin require- patients (pts) previously not controlled on OADs. Little is known how age ment in patients with T1D achieving target glycemic control with CSII or MDI might impact titration of Gla-100 and how this may affect a primary endpoint therapy under real-life condition. (PE) of a FBG ≤110 mg/dL or reaching an individual HbA1c target. TOP-1 is Methods: In this multicenter registry study, a total of 673 adults (≥18 a prospective observational study with a 12 months follow-up in pts with years old) and 285 children T1D (<18 years old) with CSII or MDI therapy T2DM attending GP offices in Germany. Pts on OADs with or without basal were enrolled and followed-up for at least one year. The total insulin dose insulin other than Gla-100 with an HbA1c of 7.5-10% whose physicians had (TDD), the total basal dose (TBD) and percentage of TBD to TDD (%TBD) were decided to use insulin Gla-100, were included in the study. Pts (n=2,462) collected after meeting the age-specific targets of HbA1c (<7.0% for adult, Therapeutics were grouped by age: <65 yrs (45.6%), 65-74 yrs (31.3%), ≥75 yrs (23.1%). <7.5% for children) at least for 6 months. 2 Clinical Diabetes/

The mean individual HbA1c targets were 6.8%, 6.9% and 7.1%, respec- Results: A total of 218 adults (age 33.6±11.1years, BMI 20.9±2.1kg/m ) and PUBLISHED ONLY tively. Elderly pts ≥75 yrs were less often male (43.8 vs. 58.7%), had a lower 68 children (age 11.1±4.3years, BMI 17.1±2.1 kg/m2) patients met the above BMI (29.7 vs. 32.0 kg/m2), a lower FBG (181.1 vs. 186.5 mg/dl) and a similar criteria, with the durations of diabetes 4.9 (2.1,9.6) years and 1.2 (1.0,3.5) HbA1c (8.5 vs. 8.6%) at baseline compared to pts <65 years. Insulin titration years, respectively. After 1 year follow-up, HbA1c were 6.23±0.51% and was performed according to Fritsche (46.7%), Davies (28.9%) or individu- 6.72±0.47% in adult and children group respectively. In adult patients, the ally (22.0%) with a trend for an increased use of the Fritsche scheme in the TDD was 0.62±0.24 IU/kg and 0.70±0.23 IU/kg in patients treated with CSII elderly (51.1% ≥75 years vs. 45.8% <65 years). The PE was equally often met and MDI (P=0.013), respectively. The TBD was 0.22±0.15 IU/kg and 0.18±0.12 across age groups being reached in 63.6, 68.2, and 66.9% of pts, respec- IU/kg in patients with CSII and MDI (P=0.022), respectively. In children tively (p=n.s.). While there was no differential response with respect to the patients, the TDD was 0.72±0.29 IU/kg and 0.89±0.25 IU/kg in patients with FBG (p=n.s. at 12 months), more elderly pts (54.7%) and more pts 65-74 years CSII and MDI (P=0.012), respectively, while the TBD was 0.35±0.15 IU/kg (54.3%) reached their individual HbA1c target than those younger (46.0%) and 0.26±0.11 IU/kg (P=0.017) in patients with CSII and MDI, respectively. at 6 and 12 months (p≤0.02 for the pairwise comparison). Pts ≥75 yrs had a %TBD was higher in CSII group than that in MDI group (44±12% vs. 34±10%, lesser drop of HbA1c (Δ1.2% vs. Δ1.5%) and FBG (Δ 53.4 mg/dl vs. Δ 64.1 mg/ p<0.001 in adults; 46±12% vs. 30±9%, p<0.001 in children). dl) at 12 months than younger pts. Confirmed, symptomatic hypoglycemia Conclusions: The results indicated that under real-life condition, the basal was documented in 0.3%, 0.3%, and 0.5% of pts (0.2%, 0.4% and 0.4% dur- insulin requirements were less than 50%, with higher rate by CSII than by MDI ing night) with increasing age with no difference between groups. This study in Chinese patients with T1D whose HbA1c reached the age-specific targets. shows that the targeted and achieved HbA1c levels varied according to age Supported By: National Health and Family Planning Commission Foundation for and while being less stringent in elderly subjects it was more frequently met. Public Welfare Industry Research (201502011); Science and Technology Planning Supported By: Sanofi Aventis Deutschland GmbH Project of Guangdong Province (2014A020212065, 2015A030401034); Sun Yat-Sen University (2007030) 2372‑PUB Real-World Titration of Insulin Glargine 100 U/mL in Patients with 2374‑PUB Type 2 Diabetes Poorly Controlled on Oral Antidiabetic Drugs Clinical Predictors of the Need for Further Treatment Escalation in JOCHEN SEUFERT, HELMUT ANDERTEN, ANJA BORCK, ANDREAS FRITSCHE, Patients with Type 2 Diabetes on Basal Insulin Therapy KATRIN PEGELOW, STEFAN PSCHERER, MARTIN PFOHL, Freiburg, Germany, MICHAEL A. NAUCK, CHRISTINA BUCHHOLZ, MELANIE KAHLE, JURIS MEIER, Hildesheim, Germany, Berlin, Germany, Tübingen, Germany, Weimar, Germany, Bochum, Germany, Bad Lauterberg, Germany Duisburg, Germany Patients with type 2 diabetes can achieve adequate glycemic control with Adding insulin glargine 100 U/mL (Gla-100) improves glycemic control in basal insulin/oral glucose-lowering agents, others need further treatment patients (pts) previously not controlled on OADs. While there have been titra- intensification. Our aim was to identify clinical characteristics that distin- tion schemes suggested, real-world titration of basal insulin is unknown. guished between these two patient populations. During one calendar year, TOP-1 is a prospective observational titration study with a 12 months FU in 213 out of 1042 consecutively hospitalized type 2-diabetic patients were pts with T2DM attending GP offices in Germany. Pts on OADs with or with- treated with basal insulin/oral agents. This led to satisfactory glucose out basal insulin other than Gla-100 with an HbA1c of 7.5-10%, whose physi- profiles in 156 patients (73.2%), for whom continuation of basal insulin cians had decided to use insulin Gla-100, were included. Pts (n=2,308) were treatment was recommended, while in 57 (26.8%), intensification of treat- grouped into titration groups: no titration (0 U; 39.2%) within the first month, ment was deemed necessary (clinical decision) and initiated. We compared titration +1-4 U (31.0%), +5-8 U (17.7%), and >8 U (12.1%). The mean individual patients’ characteristics, the insulin titration process, and resulting in- HbA1c target was 6.9% in all groups except the > 8 U group (7.0%). Pts with hospital plasma glucose profiles between these groups. Patients needing forced titration (>8 U) were younger (63.8 vs. 66.0 yrs), more often female intensified regimens (adding meal-time insulin or GLP-1 receptor agonists 2 (58.1 vs. 49.8%), had a higher BMI (33.2 vs. 30.6 kg/m ), and had a higher or switching to premixed insulin) had higher initial HbA1c values (10.1 ± 1.7 FBG (201.0 vs. 179.4 mg/dl) as well as higher HbA1c (8.8 vs. 8.4%) at base- vs. 9.5 ± 1.9%, p = 0.028), were more likely to be female (49.1 vs. 25.0%, line compared to no titration. Pts with forced titration had a steeper mean p = 0.0014) and more obese, and required more basal insulin to achieve fast- drop of HbA1c (Δ1.6% vs. Δ1.2%) and FBG (Δ 75.1 mg/dl vs. Δ 51.2 mg/dl) ing plasma glucose targets (62 ± 40 vs. 39 ± 27 IU/d, p < 0.0001. In both at 12 months than patients with no titration. The primary endpoint (FBG ≤110 patient groups, basal insulin was able to reduce fasting plasma glucose mg/dL or an individual HbA1c target) was, however, equally often met in titra- into the target range (119 ± 23 vs. 103 ± 15 mg/dl), however, in those need- tion groups with no, 1-4 U, 5-8 U and >8 U, being reached in 65.0, 68.4, 66.7, ing treatment intensification, post-prandial plasma glucose remained sub- and 62.9% of pts, respectively (p=n.s.). A combination of FBG ≤110 mg/dL stantially higher after basal insulin titration (226 ± 36 vs.169 ± 28 mg/dl, and an individual HbA1c target was met at 12 months more often in pts with p < 0.0001). After therapy intensification (i.e., before hospital discharge),

ADA-Supported Research For author disclosure information, see page A751.

A619 CLINICAL THERAPEUTICS/NEWCATEGORY TECHNOLOGY—INSULINS

similar plasma glucose profiles were reached in both groups. Those needing 2377‑PUB intensification had significantly longer hospital stays. Effects of Gender on Short-Term Intensive Insulin Therapy for Newly In conclusion, basal insulin therapy can provide satisfactory glucose con- Diagnosed Type 2 Diabetes trol in more than 70% of patients with type 2 diabetes. Long diabetes dura- XUESI WAN, ZHIMIN HUANG, LIEHUA LIU, YANBING LI, Guangzhou, China tion, obesity, insulin resistance and female sex are associated with the need It has been demonstrated that patients with newly diagnosed type 2 dia- for further treatment intensification. betes mellitus (T2DM) achieved drug-free glycaemic remission after short- term intensive Insulin treatment (IIT). It’s interesting that the male were bet- 2375‑PUB ter restored than the female. The purpose of this study is to clarify the effect The Final Dose of Basal Insulin Needed for Effective Blood Glucose and mechanism of gender on T2DM remission and related determinants. An Control and Determinants in Type 2 Diabetes Mellitus in China: observation of 160 patients with newly diagnosed T2DM (n=160,103 male, Results from Observational Registry of Basal Insulin Treatment aged 50±10 ys, HbA1c 11±2.1%) after 2 weeks of CSII was studied diabetes (ORBIT) remission rates. After one year of the termination of intensive therapy, there LEILI GAO, LINONG JI, PUHONG ZHANG, Beijing, China were 83% male and 65% female still in remission. The characteristics of Background: The efficacy of basal insulin (BI) for adequate glycemic con- male and female patients were compared in Table below. Data with nor- trol in patients with type 2 diabetes mellitus (T2DM) has been well docu- mally distributed variables showed as mean ± SD. Obviously, the ages at mented by randomized clinical trials. This post-hoc analysis of the ORBIT diagnosis of T2DM were much younger in male when compared with the study was performed to explore the final dose of BI used in insulin-naïve female (47±10 ys vs. 54 ±9 ys). And waist-to-hip ratios were higher in the T2DM patients and determine the patient characteristics that affect the male. GHbA1c and BMI in the male group seemed to be higher but without final dose of BI in the setting of real-world clinics in China. significant difference. There were no differences in DM family history (43% Methods: This multicenter observational registry enrolled 19,894 adult vs. 48%), HOMA-β, HOMA-IR and average daily insulin dose between the T2DM patients with inadequately controlled hyperglycemia and treated two groups. Moreover, the single factor logistic regression showed the age with oral antidiabetic drugs (OADs) from 209 hospitals across all 8 regions in significantly associated with T2DM remission (OR=0.95). These indicated Mainland China. Of these patients, 5191 who continued to receive BI after 6 that gender might affect the remission rate mainly by the age at diagnosis. months and achieved HbA1c target (<7%) were analyzed. Patient character- Younger patients seem to have a more favorable outcome. istics including age, body weight, fasting plasma glucose (FPG), use of OADs Table. Differences between Gender. and insulin (type and dose), glycemic control, and hypoglycemic episodes Therapeutics Male Female were recorded at baseline and 3- and 6-month follow-ups. Clinical Diabetes/ Before IIT After IIT Before IIT After IIT

PUBLISHED ONLY Results: The dose of BI needed for effective glycemic control was 0.20±0.08 IU/kg/day. High body mass index, high fasting plasma glucose, BMI(kg/m2) 25.6±3.2 25.3±3.0 24.7±3.4 24.4±3.0 young age, longer duration of diabetes or OAD treatment, more number of waist-to-hip ratios 0.95±0.05 0.95±0.05 0.89±0.06 0.88±0.06 OADs at baseline, and allocation to detemir and glargine were significant GHbA1c(%) 11.2±2.0 9.6±1.6 10.5±2.1 9.1±1.6 independent predictors for high dose of BI. Conclusion: This post-hoc analysis of the ORBIT registry provides key HOMA-β 22.5±16.6 67.2±52.0 26.3±20.8 63.0±40.8 information on the final dose of BI needed for effective glycemic control in HOMA-IR 3.9±2.2 2.1±1.2 4.3±2.7 2.2±1.1 Chinese T2DM patients. Furthermore, this study identified crucial patient average daily insulin dose 0.64±0.17 — 0.66±0.15 — characteristics that are significant determinants of the final dose of BI in a real-world setting. 2378‑PUB 2376‑PUB Physician and Patient Perspectives after Switching Current Basal Comparison of Clinical Outcomes in Inpatient Diabetes Manage‑ Insulin to Insulin Degludec in Type 1 Diabetes Mellitus (T1DM) ment by Endocrine Team and Non-Endocrine Teams in Cardiac Sur‑ Patients with Previous Hypoglycemia gery Patients EDURNE LECUMBERRI, MAITE ORTEGA, JOSÉ ANTONIO QUESADA, MARTA JAGDEESH ULLAL, DAVID A. KLIMPL, ELIAS S. SIRAJ, Norfolk, VA, Baltimore, MD ITURREGUI, ALICIA ESTRELLA, CLOTILDE VAZQUEZ, DOMINGO OROZCO-BEL- Maintaining reasonable glycemic control in Coronary Artery Bypass Graft TRAN, Madrid, Spain, Alicante, Spain (CABG) patients has proven to improve outcomes. Our Endocrine Team (ET) Introduction: Insulin analog Degludec has already demonstrated to reduce is routinely consulted in patients with diabetes mellitus (DM) undergoing severe hypoglycemic events and to slightly improve glycemic control. How- CABG where we use carbohydrate counting (CC) for prandial insulin dosing, ever, there is scarce data from real-life clinical practice studies about quality which has been shown to improve glycemic control when compared to fixed of life and fear of hypoglycemia in this group of patients. dose (FD) bolus dosing. When we are not consulted, various non-endocrine Objectives: To assess efficacy, safety and patient perception of hypogly- teams (NET) manage the diabetes, using an FD bolus regimen. cemia and quality of life after 6-month-treatment with insulin Degludec. This study was aimed at comparing key clinical outcomes and length of Material and Methods: Observational study. Variables: Efficacy (fasting stay (LOS) between the 2 groups of patients managed by ET and NET. It was glucose levels and A1c), safety (percentage of patients with hypoglycemia designed as a retrospective review of patients with DM who underwent and number of hypoglycemic episodes), assessment of degree of hypogly- CABG over a 5 year time frame. cemia fear (Hypoglycemia Fear Survey II (HFS-II)) and quality of life (visual A total of 3366 charts of patients with DM who underwent CABG were analogue scale (VAS) of EQ-5D test). Student’s t-test for paired samples was queried electronically. 1546 were excluded due to various exclusion criteria. used to evaluate changes in mean test punctuation. This resulted in 473 patients in the ET group and 1347 patients in the NET Results: 101 T1DM patients were included. Fasting glucose (mg/dl) (163.4 group. Data below are presented as mean ± standard deviation, comparing vs. 132.7; -30.7 ± DE 66.7; p=0.006) and HbA1c (%) (7.78 vs. 7.59; -0,18 ± DE the ET group vs. NET group. 0,66; p=0.000) reduction was observed. Reduction of percentage of patients The HbA1c was 8.2 ± 3 vs. 6.9 ± 2.4%. The LOS was 13.8 ± 10.9 days vs. with severe hypoglycemia (11.9 vs. 4%) and reduction in number of hypo- 14.2 ± 10.5 days. The number of hyperglycemic events >180 mg/dl were 25.8 glycemic episodes (0.17 vs. 0.05; p=0.003) were observed. After the basal ± 23.0 vs. 18.0 ± 20.2. The number of hypoglycemic events (<70 mg/dl) were insulin switch the mean punctuation of the fear of hypoglycemia test HFS-II 2.8 ± 4.3 vs. 2.1 ± 3.7. Using a multivariate analysis, increasing age, HbA1c, was significantly reduced (24.17 a 20.26; p<0.001), the mean punctuation in and female gender were all associated with a longer LOS (P < 0.05). LOS was VAS was improved in some degree as well (71,6 a 73,9; p=0.081). slightly lower (though non-significant) in the ET group, where CC was used Conclusions: Insulin Degludec improves slightly glucose control in this compared to NET where FD was used. The frequency of both hyperglycemic group of type 1 DM patients. It reduces significantly the rate of severe hypo- and hypoglycemic episodes were significantly higher in the ET group com- glycemia reported by patients. They show less fear of having a hypoglycemic pared to NET group (P <0.01). episode, this fact possibly being an indirect sign of the decrease of the fre- In summary, our study showed that HbA1c, age and gender were major quency of them. Although it does not reach statistic significance, the mean drivers of LOS. The LOS in the ET group was equal to that of the NET group, score of VAS has slightly increased. This could be seen as a sign of a modest despite the possible confounding resulting from more complicated patients improvement in the perception of health status. ending up in the ET service as demonstrated by a higher HbA1c, and more frequent hyperglycemic and hypoglycemic episodes.

For author disclosure information, see page A751. ADA-Supported Research

A620 CLINICAL THERAPEUTICS/NEW TECHNOLOGY—INSULINCATEGORY DELIVERY SYSTEMS

CLINICAL THERAPEUTICS/NEW TECHNOLOGY— 2381‑PUB INSULIN DELIVERY SYSTEMS WITHDRAWN 2379‑PUB WITHDRAWN Therapeutics Clinical Diabetes/ PUBLISHED ONLY

2380‑PUB Predictive Low Glucose System in People with Type 1 Diabetes Hard to Treat 2382‑PUB DAVIDE BRANCATO, MATTIA FLERES, GABRIELLA SAURA, VITO AIELLO, ALES- Regulation of Blood Glucose Level in Type 1 Diabetes by Micropro‑ SANDRO SCORSONE, ANNA DI NOTO, FRANCESCA PROVENZANO, LUCIA cessor-Controlled Automated Implanted Patch SPANO, VINCENZO PROVENZANO, Partinico, Italy WI KIM, SUWANG JANG, SANGKEON PARK, BUMKEUN PARK, JUNGHO YOON, Aim: To study the effect of Predictive Low Glucose System (PLGS) in peo- Daejeon, Republic of Korea ple with T1D who show persistently high levels of glycemic variability and/or Currently there have been lots of trials by controlling blood glucose level A1c, despite efforts to improve diabetes management. by using automated implanted insulin secretion system in IDDM patients. Methodology: From 120 people with T1D using PLGS, we retrospectively Most of the automated insulin secretion systems are carried out by using air selected 20 consecutive subjects (9 females, 11 males; age, mean ± SD = based or spring buttons to insert insulin and lower down blood glucose level 23.3 ± 14.7) who showed, during the last year and despite repeated and when concentration is over 120mg/dl. However, depend on the sensitivity intensive efforts to improve diabetes management: 1) ≥ 1 severe hypoglyce- and the pressure controlled by patient, insulin concentration is inaccurately mia and 2) A1c ≥ 8.0% or ≥ 1 hospital admissions for diabetes acute compli- secreted by patients. Overall object of this study was to controlling blood cations. Main outcome were, before and 3 months after the commencement glucose level to IDDM patient by inserting dome-shape insulin secreting of PLGS: glycemic variability (GV), expressed as SD from the mean glucose implanted patch onto left arm. To accomplish this, we made dome-shaped concentration measured by continuous glucose monitoring (CGM); % of time patch assembled with micro motor with 20 teeth of gear system, which below glycemia < 70 mg% (TBG), measured by CGM; and A1c. represent 1 tooth for 0.5 units, microprocessor to record and send blood Results: GV was significantly reduced from 56.2 ± 13.9 to 50.1 ± 10.2 mg% and glucose data to wireless PC or smart device, insulin capsule and sliding (p < 0.05), TBG improved from 0.37 ± 0.71 to 0.13 ± 0.23% (p < 0.05), and A1c door to secret insulin form capsule with insulin and glucose sensing crystal decreased from 8.0 ± 1.5 to 7.4 ± 0.6% (p < 0.05). No severe hypoglycemia attached at lower end. Initial In vivo examination was done by using strepto- occurred during the trimester of observation. zotocin treated C57BL/6J with surgically implanted dome-shaped patch. We Discussion and Conclusions: T1D is not an easy condition to treat and, also confirmed insulin and glucose sensing from dome-shaped patch by L6 despite the efforts to optimize management through intensive education skeletal cell and adipocyte glucose intake level. Blood glucose level of STZ programs, use of most updated insulin analogues, insulin pumps and CGM, a treated mice were in between 180mg/dl to 400mg/dl with accompanied by substantial proportion of people does not achieve treatment goals. Our pilot pancreas islet cell destruction. Concentration and timing of insulin secretion study suggests, in T1D hard to treat, that PLGS could improve GV and time was programmed into microprocessor when blood glucose level was more spent into hypoglycemia while decreasing A1c, thus allowing the achieve- than 120mg/dl and stop secreting insulin when blood glucose level reaches ment of treatment goals. 120mg/dl. One week later patch implanting surgery, blood glucose level was checked every one hour and patched mice drop down 80mg/dl to 120mg/dl. In summary, we believe real time controlling blood glucose level is more efficient than one time insulin injection. In particular, we observed patch with gear control insulin secretion accuracy was less than 0.01% concentration error after 2,000 secretions.

ADA-Supported Research For author disclosure information, see page A751.

A621 CLINICAL THERAPEUTICS/NEW TECHNOLOGY—NONINSULINCATEGORY INJECTABLES

2383‑PUB for all the needle lengths; though these data were weak. Shorter needles Insulin Pump Therapy in Patients with Diabetes Mellitus Type 2: in lean patients were associated with a low risk of intramuscular injection. A Survey of Patient Expectations Much inference has been made from this biased research. With better study FELIX ABERER, TINA POETTLER, HARALD SOURIJ, NORBERT TRIPOLT, SEBAS- design the literature may suggest that patients prefer any needle length from TIAN BECVAR, SILVIA LEITGEB, THOMAS PIEBER, EVA NOVAK, JULIA K. MADER, 4 to 8mm. It may be that short needles should be advocated in lean people; Graz, Austria however, as the majority of insulin users are obese a more patient centred Background: Reimbursement and practitioner´s recommendations for approach would be to offer a selection of needle lengths. the use of continuous subcutaneous insulin infusion (CSII) devices in type 2 diabetes (T2D) is restricted to a minority. Not only complexity of available 2385‑PUB systems but also concerns about efficacy and safety exist. Simple devices, Medtronic SmartGuard™ Technology Results in Positive Psycho‑ which do not require programming and provide easy handling, might over- logical Well-Being in People Living with Diabetes come these issues. Hereby we aimed to understand the knowledge and the SHWETA GOPALAKRISHNAN, JOHN MUECKLER, JOHN SHIN, Northridge, CA attitude towards CSII of patients with T2D. Background: Patient-reported outcome on the impact of therapy on quality Material and Methods: A questionnaire survey in T2D was performed. It of life (QoL) is an important measure of successful diabetes management. (21 questions) contained extractions from standardized questionnaires (bar- Several aspects of psychological well-being of patients using the MiniMed® riers to insulin treatment, insulin treatment appraisal, problem areas in dia- 530G system with SmartGuard technology (Threshold Suspend-TS) were betes) which focus on insulin therapy, diabetes treatment and quality of life evaluated. aspects. For this analysis answers YES, rather YES, NO and rather NO were Methods: From June 2016 to November 2016, responses were collected summarized to YES and NO. from TS users (n=406, ages 18-65 years) completing the (Bradley) Well- Results: 100 patients (50 w, 50 w/o insulin therapy) participated. being questionnaire, 3-4 months after beginning MM530G therapy. Users Patient characteristics: 29% female, mean age 63.3 y, diabetes duration included former MDI users (n=113), upgrade pumpers (n=267), and competi- 12.8 ± 9.1 y, A1c 57 ± 11 mmol/mol. While majority of subjects have never tive switchers (n=26). The questionnaire measures psychological well-being seen a CSII before and believes that it is more complicated than MDI treat- in people with chronic illnesses and has been recommended by the World ment, about 70% of them believe that it might be helpful and more than 80% Health Organization for widespread use to evaluate impact of treatment and are confident, they could handle it (Figure). overall patient health. It measures positive and negative well-being on a Conclusion: Acceptance of CSII is high in patients with T2D. Clinical trials, 4-point scale (0-3, Not at all to All the time). Therapeutics testing easy-to-use devices will be needed to evaluate broader use in T2D. Results: The Table shows percent of TS users responding to the survey Clinical Diabetes/

PUBLISHED ONLY Figure. as experiencing minimal to no psychological issues. For the most part, more than 60% of patients showed positive well-being responses and less than 20% had responses showing negative well-being 3-4 months after beginning MM 530G therapy. Table.

Supported By: Medtronic 2384‑PUB Is a 4mm Insulin Pen Needle Best for All? An Independent Review of the Literature CLINICAL THERAPEUTICS/NEW TECHNOLOGY— HENRIETTA E. MULNIER, HARIETT POPE, HAYA ABU GHAZALEH, London, United NONINSULIN INJECTABLES Kingdom Recent International advice on insulin delivery advocates the use of a 4mm 2386‑PUB insulin pen needle for all. However, the evidence base is weak and potentially Titratable Fixed-Ratio vs. Sequential Combination of Insulin biased. Here we present an independent systematic critical review. A system- Glargine and Lixisenatide: Propensity Score-Matched Analysis to atic search of MEDLINE, EMBASE, and CINAHL was undertaken in November Compare LixiLan-L and GetGoal Duo-2 Trials 2016. The search strategy combined three facets: diabetes mellitus, insulin JURIS J. MEIER, JOSEP VIDAL, MINZHI LIU, RICCARDO PERFETTI, JULIO ROSEN- pen needle, and adverse effects. Multiple terms were used to define each STOCK, Bochum, Germany, Barcelona, Spain, Somerset, NJ, Bridgewater, NJ, Dallas, concept and only English-written articles published on or after 2006 were TX included in the screening process. Free-text searches were also performed. Treatment of T2DM uncontrolled on basal insulin glargine (iGlar) may be The search extracted 2,382 citations. Sixteen articles were included in the advanced with lixisenatide (Lixi) as a separate injection or switched to a literature synthesis; 11 trials and five observational studies. Of the identified titratable fixed-ratio combination with Lixi (iGlarLixi). These approaches papers 12 (75%) were funded by insulin pen manufacturing companies. Criti- were indirectly compared by propensity score-matching according to base- cal appraisal scores were mostly low to very low, with serious risk of bias line covariates to minimize confounding; in long-standing iGlar-treated such as lack of blinding, no power calculation, or influence of study design uncontrolled T2DM, simultaneous administration with iGlarLixi in the Lixi- particularly in the use of scoring measures. Shorter needles were in general Lan-L trial (n=367) was compared with sequentially adding Lixi in GetGoal reported as preferred, though a large proportion of patients also stated no Duo-2 (n=298). In 241 well-matched pairs comprising ~80% of participants, preference or preferred the longer needle. One of the ten studies that mea- iGlarLixi provided significantly greater reductions in HbA1c, higher propor- sured pain reported less pain with a 6mm needle compared to a 5mm, and tions of patients attaining HbA <7%, and greater decreases in PPG excur- two further studies reported no difference in pain scores for 5mm compared 1c sions (Table). Despite lower HbA1c with iGlarLixi, the incidence of hypoglyce- to 8mm needles. One paper reported less pain using larger volumes and an mia was similar between groups. Rates of GI AEs were lower with iGlarLixi 8mm needle than with a 6mm. Bleeding leakage and bruising scored similarly

For author disclosure information, see page A751. ADA-Supported Research

A622 CLINICAL THERAPEUTICS/NEW TECHNOLOGY—NONINSULINCATEGORY INJECTABLES vs. sequential addition of Lixi, likely due to slower up titration with iGlarLixi. between the change of INS and change of BM at visit 4 vs. visit 1 (r = 0.582, Administration of iGlarLixi may have an intrinsic complementary mechanism P = 0.029) and also at visit 3 vs. visit 1 (r = 0.574, P = 0.032). So, incretins may of action controlling simultaneously both PPG and FPG in a more physi- be considered as an effective addition to CSII + M treated PWD2. ological way. This indirect comparison suggests that in basal insulin-treated Table 1. Diabetes Control at Visit 1-4. N=14; Median, Min-Max; P <0.05. T2DM in need of insulin intensification, iGlarLixi may offer an advantage over combining insulin and a GLP-1 RA in two separate injections. Visit 1 before CSII 2 CSII+M 3 CSII+M 4 CSII+M 2 vs. 1 P 4 vs. 1 P 4 vs. 3 P +incretin Insulin IU/d 85 60 77.3 64.3 0.022 NS NS 38-122 36.7-77 61-112 47-98 HbA1c mmol/mol 80 78 79 63 NS NS 0.008 65-91 52-118 66-117 51-109 MPG mmol/l 11.4 10.3 9.4 7.7 NS 0.008 0.004 8.8-14.5 8-12.7 6.7-12.7 5.5-12.2 Body Mass kg 110.8 108.7 111.6 109.6 NS NS 0.008 89-147.5 87.9-145.4 91.8-144.4 90.6-143.5

2388‑PUB Dulaglutide Treatment Reduces Glycemic Excursions in People with Type 2 Diabetes: A Post-hoc Analysis of Six Phase 3 Random‑ ized Clinical Trials GUILLAUME CHARPENTIER, LUIS E. GARCÍA-PÉREZ, VIVIAN T. THIEU, NAN JIA, JENNIE G. JACOBSON, VALERIA PECHTNER, Evry, France, Indianapolis, IN, Paris, France The once-weekly GLP-1 receptor agonist dulaglutide (DU) demonstrated a significant A1c reduction and the potential for weight loss in adults with

uncontrolled type 2 diabetes (T2D) in the AWARD phase 3 clinical program. Therapeutics

DU has also demonstrated significant reductions in fasting, pre- and post- Clinical Diabetes/ prandial glucose levels. To further understand the glycemic effects of DU PUBLISHED ONLY 1.5 mg and 0.75 mg, we performed a post-hoc analysis examining glycemic excursions (morning, midday, evening, and daily mean) obtained by self- monitored blood glucose testing at 6 or 12 month time points. Analysis was performed by individual study (AWARD-1, -2, -3, -6, -8 and -9), given the different background therapies. Glycemic excursions were derived by subtracting blood glucose values before each meal from blood glucose values 2 hours after the meal. Treatment with DU 1.5 mg resulted in significant reductions from baseline in daily mean glucose excursions (change: -7 to -15 mg/ dL, p<0.05) in all six trials. The effect of DU 1.5 mg on glycemic excursions was most evident in the morning followed by the evening (Table). The trend of the glycemic excursion reductions from baseline with DU 1.5 mg and DU 0.75 mg were similar. The effect of DU treatment on glycemic excursions may contribute to overall improvements in glycemic control in patients with T2D treated with DU. Table.

Supported By: Sanofi

2387‑PUB Effects of Incretins in People with Type 2 Diabetes on CSII RUDOLF CHLUP, JANA ZAPLETALOVA, TZU HSUAN CHENG, ZDENEK RAMIK, MONIKA SLEZAKOVA, HANA ZALESAKOVA, Olomouc, Czech Republic, Paseka, Czech Republic Metabolic control improvements resulting from Continuous Subcutaneous Insulin Infusion (CSII) in insulin + metformin (M) treated people with type 2 diabetes (PWD2) are often not sufficient to achieve optimal state. The pur- Supported By: Eli Lilly and Company pose of this prospective pilot study was to assess the effect of incretins (liraglutide 1.2 mg/d or exenatide QW 2 mg) added to M (3 g/d) + CSII (insulin 2389‑PUB aspart) therapy. Fourteen PWD2 (8 men) on CSII + M, without serious com- Simultaneous vs. Sequential Combination of Insulin Glargine and plications, age 52.7 (34.3-68.8) y, diabetes duration 16.5 (5.0-20.3) y, BMI 2 Lixisenatide: Propensity Score-Matched Analysis to Compare Lixi‑ 36.6 (31.6-57.6) kg/m , were monitored at 4 visits: 1) before CSII, 2) on CSII Lan-O and GetGoal Duo-1 Trials + M, 3) on CSII + M before incretin start, 4) on CSII + M + incretin after 3.1 JULIO ROSENSTOCK, F. JAVIER AMPUDIA-BLASCO, FRANCESCO GIORGINO, (2.5-7.1) months. Medians (minimum-maximum), Wilcoxon Signed-Rank test MINZHI LIU, RICCARDO PERFETTI, YEHUDA HANDELSMAN, Dallas, TX, Valencia, and Bonferroni correction were applied to assess insulin/d (INS), evolution of Spain, Bari, Italy, Somerset, NJ, Bridgewater, NJ, Tarzana, CA HbA1c, mean plasma glucose of a daily 10-point glycemic profile (MPG), body Simultaneous treatment with insulin glargine (iGlar) + lixisenatide (Lixi) mass (BM) Table 1. Correlation r (Spearman) between the change of respec- as a titratable, fixed-ratio combination (iGlarLixi) is a new treatment option tive parameter at visit 4, 3, 2 vs. visit 1 and at visit 4 vs. visit 3 was sought vs. sequential administration of each component in T2DM uncontrolled on for. P < 0.05 was considered significant. CSII appeared to enable reduction metformin ± other oral antidiabetic drugs (OADs). Propensity score matching of INS with no increase of HbA1c/MPG. Incretins added to CSII + M resulted based on baseline covariates to minimize confounding factors was used to in significant reduction of both the HbA1c and BM. Correlation was revealed indirectly compare simultaneous administration with iGlarLixi in LixiLan-O

ADA-Supported Research For author disclosure information, see page A751.

A623 CLINICAL THERAPEUTICS/NEW TECHNOLOGY—NONINSULINCATEGORY INJECTABLES

(n=469) vs. sequential initial iGlar therapy for 12 weeks, followed by addition test (MTT) was performed at the same time on day ‑1 (baseline) and day 1. of Lixi if iGlar was insufficient, in GetGoal Duo-1 (n=223). iGlarLixi showed RG7697 was safe and generally well tolerated in healthy subjects follow- significantly greater reductions in HbA1c despite lower insulin doses, greater ing s.c. injections up to 3.6 mg. Tolerability was limited by gastrointestinal- proportions of patients reaching HbA1c <7%, and greater decreases in 2-h related adverse events (nausea and vomiting) at the highest dose tested (5 PPG, FPG, and weight vs. sequential administration of iGlar and Lixi in 87 mg). RG7697 was associated with a mild increase in heart rate but did not matched pairs (Table). Rate of hypoglycemia was lower with iGlarLixi vs. provoke hypoglycemia. RG7697 plasma concentration peaked at 2 to 6 hours sequential treatment, likely due to the lower insulin dose. iGlarLixi showed post-dose and exhibited biphasic elimination with apparent terminal half- better GI tolerability, possibly due to slower Lixi titration. life in the range of 19.3 to 25.4 h. Exposure (AUC) increased with dose in a In conclusion, this indirect comparison suggests that early treatment with manner slightly greater than dose-proportionally. RG7697 had no effect on a simultaneous combination of basal insulin and GLP-1 RA may be more fasting glycemic parameters in normoglycemic volunteers. However, during effective and have better GI tolerability vs. a sequential approach of add- MTT and at doses ≥1.8 mg, it reduced the glucose Cmax (‑46.2%) and delayed ing a GLP-1 RA when basal insulin therapy needs to be advanced in T2DM glucose tmax without affecting overall exposure to glucose. RG7697 effect uncontrolled on OADs. on insulin was more pronounced with reduction of both insulin Cmax (‑64.2%) and AUC (‑50.8%). Single s.c. injections of RG7697 up to 3.6 mg were generally well tolerated. Evidence of antidiabetic effect and pharmacokinetic profiles consistent with once-daily dosing made the drug suitable to be further tested in multiple-dose clinical trials in T2D patients. Supported By: F. Hoffmann-La Roche AG; Novo Nordisk A/S

2391‑PUB The Association between iGlarLixi and Patient Satisfaction with Their Treatment’s Ability to Control Type 2 Diabetes (T2D) Is Medi‑ ated by Reduced Glycemic Variability (GV) BORIS KOVATCHEV, AUDE ROBOREL DE CLIMENS, TERRY DEX, MICHELLE ROB- ERTS, LING YANG, MINZHI LIU, LINDA GONDER-FREDERICK, Charlottesville, VA, Lyon, France, Bridgewater, NJ, Somerset, NJ Therapeutics The aim of this post-hoc analysis was to evaluate the association between Clinical Diabetes/

PUBLISHED ONLY iGlarLixi, GV, and patients’ (pts’) satisfaction with their treatment’s ability to control their diabetes (measured by the Diabetes Management [DM] score of Treatment-Related Impact Measure for Diabetes [TRIM-D].) LixiLan-L is a phase 3, 30-week RCT in pts with T2D previously on basal insulin ± OADs treated with insulin glargine 100 units/mL (iGlar) or iGlar- Lixi, a fixed-ratio combination of iGlar and the GLP-1 RA lixisenatide. The DM domain of TRIM-D contains questions on pts’ satisfaction with diabetes treatment effects. DM scores were recorded at baseline and week 30, together with 7-point self-measured plasma glucose (SMPG) profiles used to calculate several GV metrics. A DM score increase ≥ 8.2 is a clinically meaningful difference; pts with DM increase ≥ 8.2 were considered DM responders. At week 30, 39.3% of iGlarLixi vs. 32.6% of iGlar-treated pts were DM responders (P = 0.0577). Among DM responders, iGlarLixi (vs. iGlar) resulted in significantly greater reductions in mean amplitude of glucose excursions (24% vs. 5%; P < 0.0001) and area under the SMPG curve (19.1% vs. 5.6%; P < 0.0001) indicating reduced GV. Most prominent was the reduction in the magnitude of postprandial SMPG excursions (measured by the High Blood Glucose Index [HBGI]) 56% for iGlarLixi vs. 17% for iGlar, P < 0.0001. HGBI reduction was correlated with a greater change in DM score, particularly for iGlarLixi vs. iGlar (r = −0.30; P = 0.0019 vs. r = −0.19; P = 0.0822). In pts with T2D, iGlarLixi reduces GV more than iGlar alone (published data). We conclude that this differential effect is preserved in the subpopulation of pts who reported more satisfaction with their treatment’s ability to control their diabetes; this subpopulation is larger with iGlarLixi, and the self-reported effect of treatment is correlated with a reduction in the magnitude of postprandial glucose excursions. Supported By: Sanofi U.S.

Supported By: Sanofi 2392‑PUB Improvement of β-Cell Function and Cardiovascular System in Patients with Newly Diagnosed T2DM after Long-Term GLP-1R Ago‑ 2390‑PUB nist Treatment Pharmacodynamics, Pharmacokinetics, Safety, and Tolerability of YANJUN WANG, YAN CHEN, JIAXIN WANG, SHUJIE ZHAO, CHUAN ZHANG, the Novel Dual GIP/GLP-1 Agonist (RG7697) after Single Subcutane‑ Changchun, China ous Administration in Healthy Subjects Objective: To observe the effect of GLP-1 receptor agonist Exenatide AGNÈS PORTRON, SHIRIN JADIDI, NEENA SARKAR, RICHARD DIMARCHI, treatment in β-cell function and the cardiovascular system in patients with CHRISTOPHE SCHMITT, Basel, Switzerland, New York, NY, Indianapolis, IN type 2 diabetes mellitus. RG7697 (NNC0090-2746), a novel dual GIP/GLP-1 agonist for the treat- Methods: 183 patients were newly diagnosed with type 2 diabetes mel- ment of type 2 diabetes mellitus (T2D), was administered in an ascending litus, after treatment with continuous subcutaneous insulin infusion for 2-3 single-dose study as s.c. injection to evaluate safety, PK, and PD. Clinical- weeks, on the basis of diet and exercise, continue to Exenatide (5μg, twice a Trials.gov; NCT01676584. A total of 51 healthy volunteers were enrolled day) subcutaneous injection. Follow-ups were conducted on the 183 patients in this double-blind, placebo-controlled study and randomized to receive before and after treatment 1, 3, 6,12 and 24 months, measured the fasting either active drug (n=6) or placebo (n=2) in one of 7 dose cohorts (0.03-5 plasma glucose (FPG), 2h postprandial blood glucose (2hPG), glycosylated mg). Adverse events were monitored and drug concentration, fasting gly- hemoglobin (HbA1c), waist circumference, weight, body mass index (BMI), cemic parameters, vital signs, ECG, antibody formation and routine labora- Systolic blood pressure (SBP), Diastolic blood pressure (DBP), triglyceride tory parameters were measured over 72 hours post-dose. A meal tolerance (TG), total cholesterol (TC), high-density lipoprotein cholesterol (HDL-C), low

For author disclosure information, see page A751. ADA-Supported Research

A624 CLINICAL THERAPEUTICS/NEW TECHNOLOGY—NONINSULINCATEGORY INJECTABLES density lipoprotein cholesterol (LDL-C), fasting insulin (FINS), 2h postprandial (delta values from baseline (mean±SD): 0.05±0.29 vs. 0.13±0.43 nmol/nmol, insulin (INS120). Calculated insulin sensitive index (HOMA-IS) and the index P=0.45) or 8-oxodG/creatinine (0.15±0.25 vs. 0.30±0.33 nmol/nmol, P=0.10) of insulin resistance (HOMA-IR) by steady-state model assessment (HOMA). vs. no treatment. Results: All time points after treatment compared with before treatment, In conclusion, women with and without pGDM exhibit similar RNA and the changes of SBP, DBP, TG, TC, HDL-C, LDL-C, FINS, INS120, HOMA-IS, DNA oxidation, and liraglutide treatment for two years does not seem to HOMA-IR were in statistically significant (P<0.05), changes in the levels affect RNA and DNA oxidation in women with pGDM. of treatment efficacy indicators were significant at the 3-month follow-up Supported By: Novo Nordisk A/S; Novo Nordisk Foundation; AP Møller Founda- (p < 0.001); however, the levels of each indicator stabilized by the 6-month tion; Danielsens Foundation follow-up, there was no statistical difference (p > 0.05). Conclusion: Treatment with GLP-1R agonists could significantly improve 2395‑PUB β-cell function and lower blood pressure, decreases the level of blood-lipid, A Novel Dual Amylin and Calcitonin Receptor Agonist (DACRA), protect cardiovascular system in patients with newly diagnosed type 2 KBP-089, Induces Weight Loss through a Reduction In Fat, albeit diabetes. Not Lean Mass, while Improving Food Preference SOFIE GYDESEN, SARA T. HJULER, ZENIA FREVING, KIM VIETZ ANDREASSEN, 2393‑PUB NINA SONNE, MORTEN ASSER KARSDAL, KIM HENRIKSEN, Herlev, Denmark The Effects of Dulaglutide and Omarigliptin: A Crossover Trial Obesity and associated co-morbidities, such as type 2 diabetes and nonal- Assessed by Two-Week Continuous Glucose Monitoring and Treat‑ coholic fatty liver disease, are major health challenges - hence, development ment Satisfaction Survey of weight loss therapies with the ability to reduce the co-morbidities is key. TAKAHIRO TOSAKI, HIDEKI KAMIYA, TAKAMI KATAYAMA, AKEMI INAGAKI, The effect of the dual amylin and calcitonin receptor agonist (DACRA), KBP- MASAKI KONDO, ERIKO NAGAO, YUICHIRO YAMADA, YOSHIRO KATO, SHIN 089, on bodyweight, glucose homeostasis, and fatty acid accumulation in TSUNEKAWA, TATSUHITO HIMENO, SHIORI SATO, YUKI NAKAYA, JIRO NAKA- liver and muscle tissue, food preference was investigated. Further, we eluci- MURA, Nagoya, Japan, Nagakute, Japan date weight-independent effects of KBP-089 using a weight-matched group. The efficacy of dulaglutide (DU) and omarigliptin (OM) has only been High fat diet fed rats were treated with KBP-089 s.c., at 0.625, 1.25, 2.5 µg/ reported through clinical trials, and the differences between these agents kg and vehicle resulting in a dose-dependent and sustained ~17% weight under the general practice, including CGM data, are yet to be clarified. This loss by the 2.5 µg/kg. Moreover, KBP-089 reduced fat depot size and reduced study was conducted to compare the efficacy and treatment satisfaction lipid accumulation in muscle and liver. In Zucker Diabetic Fatty rats, KBP-089 (DTSQ) of DU and OM in diabetic patients. Randomization to either group A improved glucose homeostasis through improved insulin action. To obtain Therapeutics Clinical Diabetes/

or B followed after 4 wk observation (O) without drug intervention. Group A; a weight-matched group, significantly less food was offered (9% less than PUBLISHED ONLY OM 25mg for 5 wk, no OM for 4 wk, DU 0.75mg for 5 wk, and no DU for 1 wk. in the KBP-089 group). Weight-matching led to improved glucose homeo- Group B; the same pattern with OM and DU switched. In both groups, the stasis through lowered plasma insulin; however, these were inferior to the analyses of CGM were conducted during the last 7 days of O and the last 14 effect of KBP-089. In the food preference test, normal diet rats obtained days of OM or DU treatment. DTSQ was conducted and HbA1c, GA, insulin, 74% of their calories from chocolate. KBP-089 administration reduced total and C-peptide were measured in wk 4 of O and 4 wk after treatment with caloric intake, and induced a relative increase in chow consumption while each agent. (UMIN R000025287) The interim report includes 7 registered drastically lowering the chocolate compared to vehicle. The novel DACRA, subjects with 4 completions. The average glucose level was 177.8 mg/dl dur- KBP-089 induces a sustained weight loss, leading to improved metabolic ing O, unchanged with OM (5th wk: 176.0 mg/dl, 6th wk: 185.8 mg/dl), and parameters including food preference, and these are beyond those observed decreased with DU (5th wk: 148.0 mg/dl, 6th wk: 159.5 mg/dl). The variability simply by diet-induced weight loss. of glucose levels was unchanged with OM and reduced with DU. Duration with glucose levels over 140 mg/dl in 24 hrs improved from 18.7 hrs during O 2396‑PUB to 16.6 and 17.3 by OM, and 13.0 and 14.5 by DU. The total score of DTSQ pri- The Novel DACRA, KBP-089, Lowers Body Weight and AST Levels in mary factors was increased from 26.3 during O to 29.5 by OM and to 27.0 by High-Fat, High-Cholesterol Fed Rats DU. These observations suggest that glucose lowering effects of DU would SOFIE GYDESEN, MORTEN ASSER KARSDAL, KIM HENRIKSEN, Herlev, Denmark be stronger than those of OM, and that better treatment satisfaction could Obesity and associated morbidities, such as type 2 diabetes (T2D) and be obtained by OM and DU. nonalcoholic fatty liver disease (NAFLD) are among the greatest health challenges, hence there is an urgent need for treatments, which substantially 2394‑PUB reduce body weight, and improve glucose control and liver health. In this The Effect of Liraglutide on Oxidative Nucleic Acid Modifications in study, we evaluate the effect of KBP-089 - a highly potent dual amylin- and Women with Prior Gestational Diabetes Mellitus calcitonin receptor agonist (DACRA) - on body weight, glucose control, and EMIL L. LARSEN, SIGNE FOGHSGAARD, LOUISE VEDTOFTE, ELISABETH R. AST and ALT levels in rats fed a high fat, high cholesterol and cholate diet MATHIESEN, PETER DAMM, JENS A. SVARE, TINE D. CLAUSEN, TRINE HEN- (65%, 2% and 0.5%, respectively) (HFCC). 6 week old rats received high fat RIKSEN, HENRIK E. POULSEN, FILIP K. KNOP, TINA VILSBØLL, Hellerup, Denmark, diet (HFD) for 8 weeks followed by HFCC for another 8 weeks. After HFD, Copenhagen, Denmark, Herlev, Denmark, Hillerød, Denmark the rats were assigned into experimental groups according to body weight Ribonucleic acid (RNA) oxidation, measured by urinary excretion of 8-oxo- and four-step dose escalated (0.625, 1.25, 2.5 and 5.0 µg/kg, s.c., s.i.d.) once 7,8-dihydroguanosine (8-oxoGuo), is associated with mortality in patients weekly followed by 4 weeks of treatment with either dose. Dose escalation with type 2 diabetes. Animal studies have shown that liraglutide, a gluca- induced minimal and transient reduction in food intake at every escalation gon-like peptide-1 analog, lowers oxidative modifications. Here, we inves- step, and the following treatment with 1.25, 2.5 and 5 µg/kg resulted in a tigated the effect of liraglutide on oxidative nucleic acid modifications in significant ~ 8%, 16% and 17% vehicle-corrected weight loss, respectively, women with prior gestational diabetes mellitus (pGDM). Women with pGDM with a corresponding reduction in fat depots. KBP-089 treatment also led to (n=104) were randomized to one-year double-blinded intervention with lira- a reduction of the HFCC induced increase in liver weight (2.5 µg/kg, p<0.05; glutide (1.8 mg, once-daily) (n=49) or placebo (n=55) followed by one-year 5 µg/kg, p<0.01) and a significantly reduced circulation AST (1.25 µg/kg, open-label intervention. An age and BMI matched control group of healthy p<0.05; 2.5 µg/kg, p<0.05 and 5 µg/kg, p<0.01), albeit no difference was women without pGDM was also studied (n=15) at baseline. The urinary observed in ALT levels. Finally, all treatment groups showed a trend towards excretion of 8-oxoGuo and 8-oxo-7,8-dihydro-2’-deoxyguanosine (8-oxodG) lower blood glucose levels compared to vehicle during OGTT, and interest- was determined at baseline, after 53 weeks (52 weeks intervention fol- ingly, KBP-089 groups dose-dependently suppressed the glucose-induced lowed by a one-week wash-out), and after 104 weeks by ultra-performance insulin hypersecretion observed in vehicle during OGTT (1.25 µg/kg, p<0.05; liquid chromatography tandem mass-spectrometry as a measurement of 2.5 µg/kg, p<0.001 and 5 µg/kg, p<0.001). whole body RNA and deoxyribonucleic acid (DNA) oxidation, respectively. In conclusion, KBP-089 induced and sustained a significant weight loss, At baseline, similar urinary excretion of 8-oxoGuo/creatinine (P=0.86) and reduced overall adiposity, and lowered AST levels. In addition, KBP-089 8-oxodG/creatinine (P=0.39) were seen in women with and without pGDM. improved glucose tolerance, hence underscoring the potential of KBP-089 in Treatment with liraglutide for 52 weeks did not affect urinary excretion of obesity and related morbidities as T2D and NAFLD. 8-oxoGuo/creatinine (delta values (mean±SD): -0.01±0.36 vs. -0.00±0.52 nmol/nmol, P=0.91) or 8-oxodG/creatinine (-0.01±0.25 vs. 0.08±0.45 nmol/ nmol, P=0.31) vs. placebo treatment. The one-year open-label extension with liraglutide did not affect urinary excretion of 8-oxoGuo/creatinine

ADA-Supported Research For author disclosure information, see page A751.

A625 CLINICAL THERAPEUTICS/NEW TECHNOLOGY—NONINSULINCATEGORY INJECTABLES

2397‑PUB 2399‑PUB Dual Amylin and Calcitonin Receptors Agonists (DACRA) Activate WITHDRAWN Overlapping Metabolic Signaling Pathways in Muscle and Adipose Tissue ANNA T. LARSEN, SOFIE GYDESEN, KIM V. ANDREASSEN, MORTEN KARSDAL, KIM HENRIKSEN, Herlev, Denmark Dual amylin and calcitonin receptor agonists (DACRA) are novel candi- dates for treatment of type 2 diabetes and obesity due to their beneficial effects on body weight, blood glucose and insulin sensitivity. KBP-042 is a DACRA with these beneficial effects, however it is unclear how KBP- 042 mediates the improvements in metabolic status, and to what extent it involves the amylin or the calcitonin receptor. Here we searched to identify the target tissues to elucidate the effect of DACRAs directly in the tissue by evaluation of signaling pathway activation. KBP-042 mediated tissue activation was studied in vivo in Sprague Dawley rats treated with a single subcutaneous dose of the DACRA KBP-042 (5 µg/ kg) followed by collection and extraction of tissue involved in metabolism and analysis of activated signaling pathways (Erk; Akt; STAT3) by Western blot. Administration of KBP-042 (5 µg/kg) acutely activated Akt signaling in both muscle and epididymal adipose tissue in vivo. We did not observe any differences in Erk and STAT3 signaling activation in muscle and adipose tissue from rats treated with KBP-042. Furthermore, no pronounced changes in Akt, Erk and STAT3 phosphorylation were observed in other tested target tissue (liver; kidney; pancreas). In conclusion, the DACRA KBP-042 acutely activates Akt signaling in mus- Therapeutics cle and epididymal adipose tissue, indicating the beneficial metabolic effect Clinical Diabetes/ of KBP-042 directly in the metabolic tissue. Further analysis including admin- PUBLISHED ONLY istration of rat amylin or rat calcitonin will benefit to evaluate whether the improvements in metabolic status and observed tissue specific kinase activation are mediated by activation of the amylin or the calcitonin receptor.

2400‑PUB 2398‑PUB Effect of GLP-1 Analogues on Free Fatty Acid Chronic, but Not Acute, Incretin Therapy Reduces Postprandial Free KENTARO SAKAMOTO, SUMIE OKAHATA, TAKAKO MITSUMATSU, TERUO Fatty Acid Levels in Prediabetic Individuals SHIBA, Tokyo, Japan ABSALON D. GUTIERREZ, JR., KARLA BERMUDEZ SAINT ANDRE, VALA HAMIDI, Purpose: To evaluate the effect of GLP-1 receptor agonists and elucidate KAYLA RIGGS, SARA COVERDALE, AMY DURSTELER, NITYA KUMAR, HIBA ALI, the mechanisms through which to reduce glucose on the focus of free fatty VISHNU MOHAN, HEINRICH TAEGTMEYER, Houston, TX acids. GLP-1 receptor agonists and DPP-4 inhibitors exert an acute anti-inflam- Methods: We performed mixed meal tests (MMT) for patients initiating matory effect in the fasting state of type 2 diabetes mellitus (T2DM). It is GLP-1 receptor agonists (lixisenatide or liraglutide) use before and after 3 less established whether the same effects are observed in the postprandial days (3 d), and/or 1 month (1M) administration. We checked plasma glucose state of prediabetic humans. Sixteen adult prediabetic subjects (age 50 ± 2 (PG), insulin (IRI), CPR, glucagon (IRG), and free fatty acids (FFA). years, BMI 32.5 ± 0.4 kg/m2, HbA1c 5.96 ± 0.05%) participated in a single Results: 51 patients (37 men and 14 women; mean age 53.6 years and center, randomized, crossover, placebo-controlled double-blinded prospec- HbA1c 9.74±1.84%) were analyzed. The breakdown of GLP-1 receptor ago- tive trial utilizing the following medications: exenatide, saxagliptin, and nists was 26 of lixisenatide, and 25 of liraglutide. All patients underwent placebo. Fasting patients presented to the Clinical Research Unit. Baseline MMT was performed for all patients at baseline, 48 of them underwent venous blood was drawn and a single dose of study medication adminis- MMT at 3 d, and 18 at 1M. The baseline mean glucose value was (127, tered. Next, subjects ate a standardized high-fat test meal, and venous 142, 170, 204, 203, 177) at (0, 15, 30, 60, 120, 180) minutes, respectively. It blood samples were collected every 2 hours for a total of 6 hours. Blood changed to (121, 129, 145, 157, 142, 131) on 3 d, and (132, 140, 151, 170, 136, was analyzed for markers of free fatty acids (FFAs), insulin, glucose, and 126) after 1M. Whereas fasting PG values showed no significant reduction, C-reactive protein (CRP). These procedures were repeated for the remaining PG at 15 min reduced on 3 d, and PGs at 30, 60, 120, and 180 min reduced study arms. Subsequently, seven of these subjects participated in an open- after 3 d and 1M. IRI showed significant reduction at 120 and 180 min on 3 label extension study. Exenatide extended-release (ER) was administered d, and increase at fasting after 1M. CPR increased at 0, 15 min and reduced for six weeks, after which the above procedures were repeated. A single at 180 min on 3 d. IRG reduced at 30, 60, and 120 min on 3 d and increased dose of exenatide led to a smaller reduction of postprandial FFAs compared at 0, 15, and 180 min after 1M. FFA increased at 180 min on 3 d, and reduced to saxagliptin and placebo. Postprandial insulin and glucose levels were significantly at 0, 15, and 30 min after 1M. lower with single-dose exenatide, compared to saxagliptin and placebo. Six weeks of exenatide ER, compared to single dose exenatide, showed greater 2401‑PUB reductions of postprandial levels free fatty acids, as well as increased post- The Dual Amylin- and Calcitonin-Receptor Agonist KBP-042 prandial insulin and no decrease in postprandial glucose. Baseline fasting Reduces Food Intake by Activating Amlin Receptors glucose levels decreased. There were no changes in CRP levels These data SARA T. HJULER, Herlev, Denmark show that six weeks of exenatide ER significant reduced postprandial free KBP-042 is a potent peptide activator of the amylin and calcitonin recep- fatty acid levels, and effect which was not seen in acute exenatide admin- tors. In in vitro studies, KBP-042 activates both the calcitonin receptor and istration. the amylin receptor effectively. Furthermore, KBP-042 activates both recep- Supported By: National Institutes of Health (UL1TR000371); National Center for tors at a lower dose compared to the native ligands. In this study, we tested Advancing Translational Sciences (KL2TR000370) 3 different doses of calcitonin and amylin against low dose of KBP-042 in obese Sprague-Dawley rats. 6 weeks old Sprague-Dawley rats fed a high fat-diet (HFD) for ten weeks, were stratified into groups according to body weight. Study 1: Vehicle, rat Amylin 5, 50 or 500 µg/kg, rat Calcitonin 5, 50 or 500 µg/kg and KBP-042 (5 µg/kg). The rats were dosed s.c. with drug or saline and food intake and body weight were monitored. Study 2: Vehicle, rat Amylin 50 µg/kg/day, rat Calcitonin 50 µg/kg/day, KBP-042 5 µg/kg/day. All animals received subcutaneous continuous infusion (Alzet osmotic mini-

For author disclosure information, see page A751. ADA-Supported Research

A626 CLINICAL THERAPEUTICS/NEWCATEGORY TECHNOLOGY—ORAL AGENTS pump) of either saline or drug for 21 days. Food intake and body weight were 1% in 90 pts (41%) and more than 1% in 82 pts (38%) and weight less than 2 monitored. Administration of 5 µg/kg KBP-042 strongly reduced acute food Kg in 45 pts (21%), between 2-3 Kg in 83 pts (38%) and more than 3 Kg in 90 intake both after 4 hours (59%, p<0.001) and up to 24 hours (30%, p<0.001). pts (41%), mean treatment duration 14 months. The adverse events in site Reduction of food intake after 4 hours was only reduced with 500 µg/kg of injection were present for 28 pts (13%) and gastrointestinal disorders in rat calcitonin (39%, p<0.001), 50 µg/kg rat amylin (40%, p<0.001) and 500 36 pts (17%), only few patients interrupted therapy for these side effects. µg/kg rat amylin (46%, p<0.001). Similarly, in the chronic study only amylin Our results confirm efficacy and safety of exenatide once weekly in accord managed to suppress food intake significantly (34%, p<0.05 after 3 days) with literature data, although our population is older and with lower BMI whereas the food intake was unchanged using rat calcitonin. KBP-042 did compared the populations of the trials; the positive effects on HbA1c values however, have a strong effect on food intake (56%, p<0.01) and conse- and weight appear to be maintained over time but the relative short period of quently led to a reduction in body weight (6%, p<0.05). KBP-042 efficiently marketing in Italy of this drug cannot provide conclusive data. Our intention reduced body weight and improved glucose tolerance. Similar effects were is to continue observation of these patients to increase information about achieved using high doses of amylin, while the effect of rat calcitonin on durability of efficacy and drug tolerance in Italian real practice. obese rats was uncertain. These findings indicate that the primary effects of KBP-042 is mediated through activation of amylin receptors, however higher doses and mixtures of calcitonin and amylin needs to be explored. CLINICAL THERAPEUTICS/NEW TECHNOLOGY— ORAL AGENTS 2402‑PUB The Effects of Acute and Chronic Incretin Therapy on Endothelial 2404‑PUB Function in Humans with Prediabetes The Role of Dipeptidyl Peptidase-4 Inhibitors in Fat Metabolism in ABSALON D. GUTIERREZ, JR., VALA HAMIDI, KARLA BERMUDEZ SAINT ANDRE, Patients with Type 2 Diabetes SARA COVERDALE, KAYLA RIGGS, AMY DURSTELER, HIBA ALI, VISHNU MOHAN, ALEXANDER S. AMETOV, ADUATE EDUCATION STUDIES GUSENBEKOVA D.G. NITYA KUMAR, HEINRICH TAEGTMEYER, Houston, TX RUSSIAN MEDICAL ACADEMY OF POSTGR, Moscow, Russian Federation Studies show conflicting data regarding the effect in incretin thera- Objective: To evaluate the influence of combined therapy of sitagliptin and pies on endothelial function in insulin-resistant humans. We investigated metformin on fat metabolism in patients with type 2 DM. the efficacy of acute and chronic incretin therapy on postprandial endo- Materials and Methods: The study involved 82 patients with obesity, lipid thelial function and hypertriglyceridemia in prediabetic humans Fifteen

metabolism disorders, who have not reached target levels HbA1s after met- Therapeutics adult prediabetic subjects (age 51 ± 2 years, BMI 32.5 ± 0.4 kg/m2, HbA1c

formin and diet therapy. Gr 1 (n=42) received co-formulated drug, consisting Clinical Diabetes/

5.96 ± 0.05%) participated in a single center, randomized, crossover, pla- of sitagliptin 100mg and metformin 2g twice a day. Gr 2 (n=40) received PUBLISHED ONLY cebo-controlled double-blinded prospective trial utilizing the following med- metformin 2 g/day. Dynamics of FG, PPG, HbA1c, weight, WC, lipid profile, ications: exenatide, saxagliptin, and placebo. For each study arm, fasting insulin, proinsulin, leptin, adiponectin, HOMA IR, HOMA-β index was evalu- patients presented to the Clinical Research Unit. Forearm blood flow (FBF) ated at baseline and at 6 months of therapy. MRI was performed to assess was measured via strain gauge venous occlusion plethysmography. Blood visc fat in all the patients. was drawn for lipid levels. Study medication was given and subjects ate a Results: Patients in both gr demonstrated significant positive changes in standardized high-fat test meal. Blood was collected every 2 hours, and FBF the levels of FG, PPG, and HBA1c. Gr I HbA1c decreased from 8,3 to 6,6%, was measured every three hours, for a total of 6 hours. These procedures gr II from 8,35 to 7,62%. FPG and PPG in gr 1 were reduced by 2,67 and 3,26 were repeated for the remaining study arms. Subsequently, seven patients mmol/l, and gr II by 0,33 and 0,64. Gr I weight loss was 4, 9 kg, gr II 2,0 kg. participated in an open-label extension study. Exenatide extended release WC shortened by 6, 5 sm in gr I, gr II by 2,42 sm. MRI showed a significant (ER) was administered for six weeks, after which the above procedures were reduction of visceral fat area by 20,6 sm2 in gr I, compared to gr II with 5, repeated. Single doses of exenatide and sitagliptin improved postprandial 7 sm2 reduction. Dynamics of the area of the subcutaneous fat there is no hypertriglyceridemia, but showed no significant changes in resting or peak reliable dynamics between gr. No significant differences in the dynamics of postprandial FBF when compared to placebo. Six weeks of exenatide ER, TCh, HDL between the gr were found. Gr I IRI reduction was 3.45 mcU/ml, compared to single dose exenatide, improved fasting resting but not peak gr II 1.63 mcU/ml. Proinsulin level dropped down in gr I by 2, 93, in gr II by 1, FBF, but did not improve postprandial resting or peak FBF. The data show that 26, C-pept level increased by 1, 4 ng/ml, in gr II 0.16 ng/ml. HOMA β grew up chronic exenatide ER therapy led to mild improvements in fasting but not in gr I by 23.4 SU, in gr II by 4.8 SU. Proinsulin/insulin ratio dropped down in postprandial endothelial function in prediabetic humans, via FBF measure- gr I by 0, 19, in gr II by 0, 02. There were no significant differences between ments. Acute exenatide and saxagliptin did not improve endothelial func- the groups in the dynamics of HOMA IR and both gr demonstrated positive tion, though postprandial triglyceride levels were reduced with both drugs. dynamics. Adiponectin levels were different between the gr there was an Long-term administration of exenatide may offer cardiovascular protection increase by 1.9 ng/ml in gr I, in gr II by 0.49 ng/ml. Leptin lowered by 7.37 ng/ beyond glucose control in prediabetes. ml in gr I, in gr II by 1.21 ng/ml. Supported By: National Institutes of Health (UL1TR000371); National Center for Conclusion: Combined therapy sitagliptin and metformin received glyce- Advancing Translational Research (KL2TR000370) mic important and nonglikemic effects and noted improvement in function of B cells of the pancreas. 2403‑PUB Efficacy and Safety of Exenatide Once Weekly in Italian Real 2405‑PUB Practice Independent Predictors for HbA1c Decline with Sitagliptin as a OLGA E. DISOTEO, PAOLO ERPOLI, LAURA BARUFFALDI, ENRICA CHEBAT, MARIA Third Line Oral Hypoglycemic Agent in Poorly Controlled Type 2 E. MALIGHETTI, RAFFAELLA MANNA, PAOLO MARENCO, ANTONIO ROSSI, Diabetes Patients in a Singapore Primary Care Setting LAURA SALI, Milan, Italy, Gallarate, Italy, Cesano Boscone, Italy, Tradate, Italy, POH CHING TAN, WEI LIANG DAVID NG, KHAI WEI TAN, YONGTAI GARETH YEO, Garbagnate Milanese, Italy, Saronno, Italy YILIN JIANG, Singapore, Singapore Exenatide once weekly, a glucagon-like peptide-1 receptor agonist, Background: There is uncertainty whether dipeptidyl peptidase-4 inhibi- has shown to improve glycemic control and weight loss in type 2 diabetic tors (e.g., sitagliptin) are equally effective in different groups of patients patients. These positive effects in randomized clinical trials were lasting with poorly controlled type 2 diabetes on two oral hypoglycemic agents. and there was no evidence of loss of these effects while exenatide once Our study aimed to quantify the improvement in HbA1c when sitagliptin was weekly treatment was continued. To evaluate if these effects are pres- started as a third line agent at varying HbA1c levels, and to identify the ent also in our real practice we conducted a retrospective review of 218 independent predictors of sitagliptin efficacy in these patients. type 2 diabetic patients (pts) started therapy with exenatide once weekly Method: The clinical records of type 2 diabetes patients on maximal doses from March 2014 to June 2016 in our Diabetes Units. At the beginning of of metformin and glipizide, and newly started on sitagliptin with baseline therapy all pts had HbA1c between 7,5% (58 mmol/mol) and 8,5% (69 mmol/ HbA1c between 7 to 8.9% were retrospectively compared with a matched mol) in accordance with refund indications of our National Health Service, cohort of patients with baseline HbA1c ≥ 9%. The primary outcome was the age less than 50 years 45 pts (21%), 50-70 yrs 142 pts (65%), more than 70 yrs change in HbA1c (∆ HbA1c) after 6 months of adding sitagliptin as a third line 31 pts (14%), BMI less than 25 Kg/m2 7 pts (3%), between 25-30Kg/m2 85 pts agent. Independent predictors of greater HbA1c reduction were identified (39%), more than 30 Kg/m2 126 pts (58%), 140 pts (64%) were in dual therapy using univariate and multivariate analysis. and 112 pts were female (51%). Our data demonstrated that exenatide once Results: 327 patients were included in the study. ∆ HbA1c in higher base- weekly decreases HbA1c less than 0,5% in 46 pts (21%), between 0,5% and line HbA1c patients (≥9%) was significantly greater than the lower base-

ADA-Supported Research For author disclosure information, see page A751.

A627 CLINICAL THERAPEUTICS/NEWCATEGORY TECHNOLOGY—ORAL AGENTS

line HbA1c (7- 8.9%) patients (1.10 vs. 0.60; p<0.001). Patients who were and 0.22±0.03, n=2727 R2=0.02, INT=0.5% in MET. The slope of H relative to acarbose-naïve before sitagliptin initiation had a greater ∆ HbA1c, as com- TG was 0.55±0.05, n=1613 R2=0.13, INT= −0.03 in DN and 0.67±0.04, n=1904 pared to those who had used acarbose at baseline (0.90 vs. 0.50; p< 0.001). R2=0.18, INT=−2.60 in MET. There was thus a much more robust increase in H Multivariate analysis showed that independent predictors of greater HbA1c (0.65 per % HbA1c) with increasing baseline HbA1c in the MET treated group improvement after adjusting for potential confounders were higher baseline than in the DN group (0.10 per % HbA1c). Both groups have an H of 5.6 at HbA1c HbA1c, older age and being acarbose-naïve. There was no significant differ- of 9.9%. In the DN patients at an HbA1c of 6% the H is 5.2, so increasing HbA1c ence in HbA1c reduction between patients on sitagliptin 50mg and 100mg up to 10% is only increasing IR ~7%. On the other hand, treating patients with (0.80 vs. 0.80; p= 0.546). MET from an HbA1c of 10% to 8% is dropping H to 4.4 which is a decrease of Conclusion: Our analysis showed that sitagliptin, as a third line hypoglyce- 22% and it would be decreased by 46% down to HbA1c of 6%. There was also mic agent showed significant glycemic improvement in patients with higher an increase in TG (0.22 per % HbA1c) with increasing baseline HbA1c in the baseline HbA1c, older age and being acarbose-naïve. Analysis of a larger popu- MET than in DN (0.08 per % HbA1c). Both groups have a TG of 2.25 at HbA1c of lation over an extended period of time is warranted to confirm these findings. 8.3%. In the DN patients at an HbA1c of 6% the TG is 2.07, so increasing HbA1c up to 8.3% is only increasing TG about 8.7%. On the other hand, reducing MET 2406‑PUB from an HbA1c of 8.3% to 6% is dropping TG to 1.75 which is a decrease of Baseline HbA1c Level Correlates with Vildagliptin’s Efficacy for 22%. Furthermore, H is better correlated with TG in MET than DN. This study T2DM Treatment: A Subgroup Analysis of VISION Study therefore shows that the relationship between TG or IR and HbA1c is different LI NONG JI, LING LI ZHOU, CHANG YU PAN, JU MING LU, LE WANG, BIN HUI in DN and MET. This indicates that MET reduces TG and IR to a greater extent WANG, JAMES HE, Beijing, China, Shanghai, China than would be predicted based upon its effect to reduce HbA1c per se. In con- Correlation between baseline HbA1c and efficacy of Chinese patients trast to TG, it is surprising that H is independent of HbA1c. with T2DM was assessed using data from the fully published VISION study Supported By: Novartis (ClinicalTrials.gov: NCT01541956). The post-hoc retrospective analysis of the VISION data was conducted on the full analysis set (FAS) population with last 2408‑PUB observation carried forward (LOCF). Patients receiving Vildagliptin plus low Characteristics of Patients Initiating Empagliflozin or Other Nonin‑ dose metformin (VLDM) and high-dose metformin (HDM) groups were fur- sulin Glucose Lowering Drugs (GLDs) in the United Kingdom (UK) therly grouped into 3 tertiles/subgroups based on their baseline HbA1c levels: SOULMAZ FAZELI FARSANI, CHRISTINA RAABE, CYNTHIA J. GIRMAN, WILL “HbA1c ≤ 6.7%” (1st tertile), “6.7 7.4%” (3rd tertile). HbA1c changes from baseline (ΔHbA1c) at weeks 12 and Ingelheim, Germany, Chapel Hill, NC, Berkshire, United Kingdom, Ridgefield, CT Clinical Diabetes/

PUBLISHED ONLY 24 were assessed in all subgroups. Our results confirmed the higher baseline Preferential prescribing of the different noninsulin GLDs can be expected HbA1c can get more HbA1c reduction in both treatment groups. In VLDM based on their effectiveness and safety profiles. Assessing characteristics group, baseline HbA1c >7.4% was associated with significantly greater HbA1c of patients prescribed GLDs in the real-world can provide important informa- reduction vs. HDM both at week 12 (ΔHbA1c= -0.87± 0.97% vs. -0.53± 1.15%, tion to aid the conduct and interpretation of future comparative effective- P=0.001) and at week 24 (ΔHbA1c= -0.91± 1.11% vs. -0.66± 1.05%, P=0.0127). ness and safety studies. The UK Clinical Practice Research Datalink (CPRD), While patients in other two tertiles receiving VLDM also got the superior a primary case database, was used to assess baseline characteristics of HbA1c reduction vs. HDM at 12 week (P<0.001 and P=0.0185 in 2nd and 1st adults with T2D initiating a noninsulin GLD between Aug 2014 and Sep 2015 tertile respective). The results indicate the early combination therapy by acting (1st yr of empagliflozin (EMPA) availability in the UK). Of 25,928 T2D patients, on different pathological T2DM-related pathways such as VLDM can get more 110 started EMPA, 3,407 other SGLT2i, and 22,411 other GLDs. EMPA (mean: HbA1c reduction, especially for patients with higher baseline HbA1c. 56.8 years) and GLP-1a (57.6) users were younger than users of other GLDs (range: 61.0-65.0). Mean HbA1c for EMPA users (77.3 mmol/mol) was lower Figure. than for GLP-1a (80.7) and other SGLT2i (79.5) but higher for DPP-4i (73.8) and metformin (70.3) users. More patients starting EMPA, other SGLT2i, or other GLDs (range: 33.6-34.8%) had baseline diabetes complications vs. metformin (13.1%) or SU users (24.4%) which could may be explained by the shorter diabetes duration in metformin and SU initiators. Eye complications (23.6%) and hypoglycemia (10.0%) were the most prevalent diabetes mani- festations in EMPA users. Mean BMI in EMPA (35.4 kg/m2), other SGLT2i (35.2), and GLP-1a (37.7) users were higher than in DPP-4i, metformin, and SU users (range: 31.6-33.0). Mean eGFR of EMPA (95.5 ml/min) and other SGLT2i users (93.4) were higher than other groups; the lowest mean eGFR was observed for DPP-4i (80.7) and SU users (85.0). In the 1st year of availability, EMPA initiators were generally younger with higher BMI and eGFR, and a higher frequency of hypoglycemia history vs. initiators of other GLDs. These early patterns of prescribing for EMPA differ from other GLDs, including other SGLT2i, need to be considered in future comparative studies and the interpretation of spontaneous safety reporting.

2409‑PUB Is the Combo of Sulfonylurea plus Metformin Appropriate for First- Line Therapy of People Newly Diagnosed with Type 2 Diabetes and with HbA1c ≥7.6% and ≤9%? Supported By: Novartis Pharmaceuticals China GUILLAUME CHARPENTIER, Corbeil-Essonnes, France In 2016, dual therapy is recommended in people with type 2 diabetes (T2D), when HbA1 is 7.5% and 9%. We conducted from 2012 to 2014 a 2407‑PUB c ≥ ≤ multinational multicentre open-label 3-arm superiority trial in Africa, Middle Fasting Triglyceride and HOMA-IR Values Are Differently Related East, Russia and Latin America to compare the combination of glimepiride to HbA1c in Drug-Naïve vs. Metformin-Treated Patients with T2DM: + metformin (G+M) vs. glimepiride (G) and metformin (M) alone in newly An Analysis in a Large Pool of Patients diagnosed T2D patients. A total of 541 were randomized (182 in the G, 177 JAMES E. FOLEY, BO AHRÉN, East Hanover, NJ, Lund, Sweden in the M and 182 in the G+M arm). At baseline, mean age was 50.9±10.0 Fasting Triglyceride (TG) and HOMA-IR (H) values are surrogates of insulin years, mean BMI 29.1±4.1 kg/m2, median diabetes duration 2.3 months, resistance (IR). The current analysis was undertaken in order to assess the mean HbA1c 8.2±0.4%. Dose adjustment aimed to achieve a self-monitored impact of TG and H on HbA1c values in drug-naïve (DN) or metformin (MET) plasma glucose level >70mg/dL and 130mg/dL without symptomatic treated patients with T2DM. The baseline data of DN and of MET in the pooled ≤ hypoglycemia. The maximum daily dose was 4mg, 1000mg and 4/2000mg vildagliptin database were utilized to calculate the slope of H and TG vs. in the G, M and G+M arms, respectively. The adjusted mean change (stan- HbA1c. The slope of H relative to HbA1c was 0.09±0.13, n=1613 R2=0.07, inter- dard error) in HbA1c at month 6 was respectively -1.36 (0.07), -1.37 (0.07) cept (INT)=4.6% in DN and 0.65±0.11, n=1904 R2=0.05, INT=−0.8% in MET. The and -1.50 (0.07) in the G, M and G+M arms, with non-significant difference slope of TG relative to HbA1c was 0.08±0.03, n=2445 R2=0.01, INT= 1.6% in DN

For author disclosure information, see page A751. ADA-Supported Research

A628 CLINICAL THERAPEUTICS/NEWCATEGORY TECHNOLOGY—ORAL AGENTS between arms. The percentage of patients with HbA1c<7% was respectively baseline) was correlated with duration of diabetes, AST and ALT at base- 64.6%, 59.3% and 71.1% and the percentage of patients with HbA1c<6.5% line, ΔFPG, Δglucagon, ΔHOMA-R, ΔSUIT index, ΔAST and ΔALT. ΔGA was 31.5%, 30.8% and 45.7%. At study end, the rate of weight decrease >5% also correlated with diabetic retinopathy, AST and ALT at baseline, ΔFPG, was 3.4% (n=6), 8.2% (n=14) and 5.7% (n=10) in the G, M and G+M arms, Δglucagon, ΔSUIT index, ΔAST and ΔALT. BIA revealed that ΔGA was cor- respectively and the rate of weight increase >5% 3.9% (n=7), 1.2% (n=2) and related not with ΔLBM but with ΔBFM. 8.0% (n=14), respectively. Frequency of hypoglycemia episodes (any type) Conclusion: The present study showed that the improvement of glycemic was mainly reported in patients receiving G. Only 5.1% of patients treated control was correlated with liver function, reduction of glucagon and body with M alone reported hypoglycemia. Severe hypoglycemia were reported fat mass. A combination therapy with liraglutide and sitagliptin can be effec- in 5 patients (0.9%): 1 patient (0.5%) in the G and 4 patients (2.2%) in the tive option for T2DM. G+M arm. These results suggest that G+M is effective with a low risk of severe hypoglycemia. 2412‑PUB Supported By: Sanofi Difference of Endogenous Insulin Response between Mixed-Meal Tolerance Test and Glucagon Stress Test Is Associated with Thera‑ 2410‑PUB peutic Effect of Incretin Related Drugs Derangement of Hepatic Metabolism Induced by Chronic Treatment HIKARU TAKAMINE, YURIA NAMIKI, HIROTO SASAKI, YUYA TAKANO, KOJI INA- of Glucokinase Activator Contributes to Its Declining Efficacy ZUMI, YUKO MUROHASHI, URU NEZU OSADA, Yokohama, Japan YOSHINORI TSUMURA, YU TSUSHIMA, AZUSA TAMURA, MAKIKO HASEBE, Mixed meal tolerance test (MMT) and glucagon stress test (GST) are use- TSUNEFUMI KOBAYASHI, Tokyo, Japan ful tests in evaluating the endogenous insulin secretion in type 2 diabetes Glucokinase (GK) in hepatocytes and pancreatic β-cells plays a key role mellitus (T2DM). We often experience the greater C peptide response (CPR- in glucose homeostasis, and a number of GK activators (GKAs) have con- R) in MMT compared to that in GST. We assumed that this difference might ducted clinical trials as therapeutic agents for type 2 diabetes. However, partly reflect the endogenous incretin effect and be associated with thera- MK-0941 (MK) and other GKAs discontinued the development because of peutic effect of incretin related drugs (IRDs). Here, we investigated the asso- declining efficacy during chronic treatment or other issues. In our nonclini- ciation between the inter-test difference in CPR-R and IRDs’ therapeutic cal study, the declining efficacy was also seen after more than 8 weeks of effect. Based on our inpatient database from April 2015 to October 2016, we administration of MK in Goto-Kakizaki rats, while it was not in TMG-123 retrospectively extracted T2DM patients started on IRDs including DPP-4 (TMG), another GKA (ADA2014#1041-P). Here, we investigated factors that inhibitors and GLP-1 receptor agonists. As an indicator of the endogenous Therapeutics make the difference in the glycemic durability between MK and TMG using incretin effect, we defined⊿ CPR-R by subtracting the CPR-R in GST from Clinical Diabetes/

Goto-Kakizaki rats. First, we evaluated the effects on pancreatic β-cell that in MMT, and analyzed its effect on the changes in HbA1c at the timing PUBLISHED ONLY function, but chronic treatment of MK did not affect plasma insulin level of 1 to 3 months after IRDs initiation. We also assessed the change in fasting and β-cell mass. Next, the effects of treatment on hepatic glucose uptake plasma glucose (FPG) during the admission period. As the result, 89 patients were measured by the isolated perfused liver. After single administration, were started on IRDs (male: 64%, mean age: 63 years old, mean HbA1c: the glucose uptakes were increased by both MK and TMG. However, after 10.0%). As the primary analysis, ⊿CPR-R was positively associated with the the chronic administration, glucose uptake in only MK-treated group was change in HbA1c although not statistically significant (r=0.21, p=0.2). In the decreased compared to the control group. At this time, protein expression secondary analysis, ⊿CPR-R showed the significant and positive association level of GK was decreased and G6Pase activity was increased in the liver of with the change in FPG (r=0.46, p<0.0001). In next, in multivariate analysis MK-treated group, but not in TMG-treated group. Thus, the declining effi- adjucted with age, duration of diabetes, BMI, eGFR, CVR-R, and concomi- cacy of MK in Goto-Kakizaki rats is attributed to the derangement of hepatic tant use of α-glucosidase inhibitors again showed the positive association glucose metabolism and not associated with pancreatic β-cell dysfunction. between ⊿CPR-R and the change in FPG (β=19.7, SE=4.6, p<0.0001). In the liver perfusion study, increased production of lactate and CO2 were In summary, our results indicated that the patients with reduced incretin found by single administration of MK, but not by TMG. The hepatic glucose effect might be IRDs responder. flux analysis using 13C isotopic tracer indicated the increasing tendency of glucose uptake by both agents but the decreasing tendency of glucose 2413‑PUB release by only MK treatment. Therefore, the pharmacological effects on Teneligliptin, a DPP-4 Inhibitor, Decreases Plasma Levels of Three hepatic glucose metabolism may be different between MK and TMG. This Chemokines (CCL11, CCL22, and CXCL10) during a Standardized difference may possibly relate to the different result in glycemic durability. Meal Test in People with Type 2 Diabetes, but Differently Regulates These Chemokines Respectively 2411‑PUB MASATO KASE, IV, TAKANORI TOMOTSUNE, SHINTARO SAKURAI, KAZU- A Combination Therapy with Liraglutide and Sitagliptin Improves NORI YANAGI, TERUO JOJIMA, TOSHIE IIJIMA, KUNIHIRO SUZUKI, KUNIHIRO Glycemic Control in Japanese Patients with Type 2 Diabetes SUZUKI, YOSHIMASA ASO, Mibu, Japan Mellitus DPP-4 rapidly cleaves not only incretin hormones but also a number of YUKI MATSUHASHI, SHINJI CHIKAZAWA, SHOU OSONOI, SATORU MIZUSHIRI, chemokines, and influences the function of chemokines. We investigated MASATO YAMAICHI, HIDEYUKI OTAKA, KAZUHISA TAKAHASHI, AYA KANBA, effects of teneligliptin (a DPP-4 inhibitor) on plasma levels of CCL11 (eotaxin), KOKI MATSUMURA, KOTA MATSUKI, ERI SATO, JUTARO TANABE, MIYUKI CCL22 (macrophage-derived chemokine; MDC), and CXCL10 (IP-10) during a YANAGIMACHI, HIROSHI MURAKAMI, MAKOTO DAIMON, Hirosaki, Japan meal test before and after 24-week treatment with teneligliptin in people Background: Liraglutide (a glucagon-like peptide-1 analogue) and sita- with type 2 diabetes. We studied 10 consecutive patients with type 2 dia- gliptin (a highly-selective dipeptidyl peptide-4 inhibitor) are widely used in betes controlled inadequately with metformin and/or sulfonylurea. A stan- combination with various kinds of other oral antidiabetic drugs and insulin dardized meal were administered at baseline and 24 weeks after treatment in Japan. We evaluated the efficacy of sitagliptin when added to liraglutide. with 20 mg/d teneligliptin. Blood was sampled at 0, 30, 60, and 120 min after Methods: Japanese patients with uncontrolled type 2 diabetes mellitus meal ingestion. We measured serum DPP-4 enzymatic activity and soluble (T2DM) (HbA1c>7.0%) who were treated with 0.9 mg of liraglutide subcu- DPP-4 antigen in addition to plasma three chemokines levels. At baseline, taneously once daily (maximum dose in Japan) were enrolled in this study plasma levels of all three chemokines were significantly decreased at 60 (n=17). The serum levels of HbA1c, glycated albumin (GA), fasting plasma min after a meal digestion. Either DPP-4 activity or soluble DPP-4 antigen glucose (FPG), insulin (IRI), C-peptide (CPR) and glucagon were measured. did not change during a meal test. Treatment with teneligliptin decreased Index of insulin resistance (HOMA-R) and insulin secretion (SUIT index) HbA1c from 8.4±1.1% at baseline to 6.8±0.5% at 24 weeks. Teneligliptin were calculated using following equations; HOMA-R=IRI*FPG/405, SUIT inhibited fasting serum DPP-4 activity by 90.13% compared with baseline. At index=1485*CPR/(FPG-61.8). Body weight (BW) and compositions (body fat 24 weeks after treatment with teneligliptin, plasma CXCL10, but not CCL11 mass; BFM, lean body mass; LBM) were analysed using the bioelectrical or CCL22, significantly decreased at 60 min after a meal test. Unexpectedly, impedance analysis (BIA). 50 mg of sitagliptin was added without alter- plasma levels of three chemokines were not increased after 24-week treating previous therapy. Differences of each parameter between baseline ment with teneligliptin, were instead significantly decreased at every point and those 12 weeks after were calculated. Results were compared using during a meal test. Changes in fasting DPP-4 activity correlated positively repeated-measure ANOVA and simple regression analysis. P<0.05 was with plasma levels of CCL22, but not CXCL10 or CCL11, at every point during a defined as statistically significant. test meal. The counts of eosinophils significantly decreased after treatment Result: HbA1c and GA were significantly decreased (p=0.025 and p=0.047, with teneligliptin. respectively). Simple regression analysis showed that ΔHbA1c (12 weeks-

ADA-Supported Research For author disclosure information, see page A751.

A629 CLINICAL THERAPEUTICS/NEWCATEGORY TECHNOLOGY—ORAL AGENTS

In conclusion, DPP-4 inhibitors may affect chemoattractant of specific 2416‑PUB types of leukocytes by decreasing plasma levels of some chemokines. (UMIN000012508). WITHDRAWN

2414‑PUB WITHDRAWN Therapeutics Clinical Diabetes/ PUBLISHED ONLY

2417‑PUB WITHDRAWN

2415‑PUB Single-Dose Pharmacokinetics and Pharmacodynamics of Sita‑ gliptin, an Inhibitor of Dipeptidyl Peptidase-4, in Healthy Indian Male Subjects GANESH V. SANGLE, MOHAN PATIL, NITIN DESHMUKH, SHANTIBHUSHAN KAMBLE, KIRAN KUMAR VUPPALAVANCHU, SUSHIL KALE, LAYEEQ BAIG, GEETCHANDRA SINGH, JAVED SHAIKH, JITENDRA TRIPATHI, ARAVINDABABU P., Aurangabad, India Sitagliptin, a dipeptidyl peptidase (DPP)-4 inhibitor approved for the management of type 2 diabetes, is reported to be more efficacious in the Indian population compared with the other ethnic groups. The aim of this study was to unravel the basis of higher efficacy by assessing the pharmacokinetics and pharmacodynamics (PK/PD) profile of single-dose sitagliptin 100 mg (Januvia) in healthy Indian male subjects. In this single-centre, single-dose, cross-over, analyst-blind study, 18 healthy subjects received single dose of sitagliptin 100 mg under fasted and fed conditions. PK/PD parameters under fasting condition were analysed.

PK parameters (Cmax, Tmax, t1/2, AUC0-∞, 0-t) were determined using Phoenix WinNonlin software. PD parameters [DPP-4 inhibition, active glucagon-like peptide-1 (GLP-1) and insulin] were determined using established methods. A total of 18 subjects were enrolled [mean age, 22.4 y (range, 18-34); mean weight, 62.7 Kg (range, 51.1 - 77.0)] in the study. PK parameters expressed in

mean (SD) are Cmax 491.7 (135.9)ng/mL; Tmax 2.9 (1.0)h, t1/2 10.4 (3.0)h, AUC0-∞ 4256.1 (509.9) and AUC0-t 4198.9 (500.6)ng.h/mL. The weighted average plasma DPP-4 inhibition over 24 h for the dose of 100 mg was ≥80%. At 2 h after a standard meal, mean weighted average of plasma active GLP-1 (geometric mean ratio) was 11.1 (9.9) pM/L in the sitagliptin group which is two-four fold higher compared to that reported in other populations. The mean weighted average (SD) plasma insulin levels were 29.9 (12.3) µIU/mL. 2418‑PUB No statistically significant differences were observed in PK/PD parameters Subgroup Analysis to Evaluate Effect of Baseline Characteristics under fed condition. [In this abstract only sitagliptin arm data was included] on Markers of Renal and Cardiovascular Risk in Type 2 Diabetes No statistically significant differences in PK and DPP-4 inhibition param- Patients Treated with SGLT2 Inhibitor (Canagliflozin) eters were observed for Indian subjects compared to that reported in non- VISHAL GUPTA, VAISHALI TELI, Mumbai, India Indian subjects. However, observed higher GLP-1 levels may correlate with Objective: To compare reduction in hs-crp (Highly selective c-reactive pro- enhanced efficacy of sitagliptin in Indian population. tein) and serum Creatinine with respect to gender, age, baseline HbA1c, BMI and duration of diabetes.

For author disclosure information, see page A751. ADA-Supported Research

A630 CLINICAL THERAPEUTICS/NEWCATEGORY TECHNOLOGY—ORAL AGENTS

Methodology: Fifty patients treated with SGLT2 inhibitor (Canaglizlozin) generally at goal with minimal highs or lows. However, despite detailed dis- were followed-up at months 3, 6 and 12 to evaluate change in body mass charge instructions on who to call for problems and follow-up, this informa- index (BMI), hs-CRP, lipid profile, creatinine and renal albumin excretion. At tion appears buried and unread in the voluminous (up to >10 pages) standard the end of one year, significant reduction was observed in HbA1c, hs-CRP discharge instructions patients receive. and serum creatinine. In the present study, we assessed effect of baseline characters of age (18-40, 41-60,> 61 years), body mass index (<23, 23-27, 2420‑PUB >27 kg/m2), HbA1c (<8 or >8%), duration of diabetes (0-5, 5-10,>10 years) The Clinical Efficacy and Safety of SGLT2 Inhibitors in Japanese on reduction in hs-CRP and Creatinine. Data was represented as Mean + 2 Patients with Type 2 Diabetes for Two Years Standard Deviation with 95% confidence interval and a p value of 0.01 or TAKAHIRO TOSAKI, HIDEKI KAMIYA, AKEMI INAGAKI, MASAKI KONDO, ERIKO lower was considered significant. Unpaired t test and Analysis of Variance NAGAO, YUICHIRO YAMADA, YOSHIRO KATO, SHIN TSUNEKAWA, TATSUHITO (ANOVA) was used to compare two and three sub-groups of the baseline HIMENO, TOMOYO HAYASAKI, SHIORI SATO, YUKI NAKAYA, JIRO NAKAMURA, characters respectively. Nagoya, Japan, Nagakute, Japan Results: Serum creatinine was significantly reduced in the male popu- To evaluate its efficacy and adverse effects, one of SGLT2 inhibitors lation as compared to their female counterparts (p value: 0.0025). There (including ipragliflozin 118, dapagliflozin 49, luseogliflozin 54, tofogliflozin 36, was a uniform reduction observed across all sg groups (p value: 0.188764), canagliflozin 70, or empagliflozin 35) was administered to 362 outpatients BMI (p value: 0.1974), duration of diabetes (p value: 0.135963) and baseline with type 2 diabetes mellitus (mean age 51.8 ± 12.0 yr, diabetes duration HbA1c levels (p value: 0.3636). There was also a uniform reduction in hsCRP 7.8 ± 7.0 yr, BMI 29.0 ± 4.9) with or without other antidiabetic agents for observed across both genders (p value: 0.2519), all age groups (p value- 24 months. 0.1887), BMI (p value: 0.2839), duration of diabetes (p value: 0.83641) and Average observation period was 11.7 months. The mean HbA1c level sig- baseline HbA1c levels (p value: 0.172778). nificantly improved from 7.61 ± 1.41% to 7.00 ± 0.86% at month 12 (n=212, Conclusion: Male gender was associated with significant greater reduc- p<0.001), but the effect was not sustained to 7.16 ± 1.13% at month 24 (n=89, tion in serum creatinine at the end of 12 weeks. SGLT2 inhibitor (Canagl- p=0.006). On the other hand, body weight was significantly reduced from izlozin) was observed to reduce Serum creatinine and hsCRP irrespective of 78.1 ± 15.3 kg to 75.1 ± 13.4 kg at month 12 (n=212, p=0.017) and furthermore age, baseline BMI, HbA1c, duration of diabetes. reduced to 74.2 ± 14.0 kg at month 24 (n=89, p<0.028). Systolic blood pressure was significantly ameliorated from 134.7 ± 16.0 mmHg to 128.1 ± 13.3 2419‑PUB at month 12 (n=214, p<0.001), and the effect was sustained to 128.5 ± 12.8 Survey on Transition from Inpatient to Outpatient for Diabetes at month 24 (n=89, p<0.001). Heart rate, estimated visceral fat area, waist Therapeutics Clinical Diabetes/

Patients Not on Insulin circumference and uric acid also improved significantly. Urinary albumin/ PUBLISHED ONLY EUNICE CHUANG, JULIE KIM, HEIDEMARIE W. MACMASTER, JOSEPH S. creatinine ratio and eGFR did not show remarkable changes. A significant CHANG, ROBERT J. RUSHAKOFF, San Francisco, CA, Los Angeles, CA, Berkeley, CA increase was observed in serum HDL cholesterol from 49.4 ± 12.7 to 53.4 Inpatient hyperglycemia is generally treated with insulin. In a previous ± 15.7 at month 24 (n=78, p=0.018). Serum alanine aminotransferase was survey, we found patients discharged to home on insulin had received appro- significantly improved from 37.4 ± 28.3 U/L to 30.0 ± 22.0 at month 12 (n=192, priate education and did well at home. However, little is known about the p=0.002). This amelioration was not related to the improvement of HbA1c, diabetes-specific, posthospital discharge issues that may arise for patients but the reduction of body weight. sent home on oral hypoglycemic agents and non-insulin injectables. A total of 48 adverse events were noted, including non-severe hypoglyce- We designed and conducted a comprehensive survey (2-4 weeks post-dis- mia (n=14), pudendal pruritus (n=4), urinary frequency (n=7), a strong hunger charge by telephone or online) covering predischarge issues such as survival sensation (n=3), nausea (n=2), constipation (n=2), itching sensation (n=2), skill education and discharge instructions, postdischarge logistical problems lightheadedness after alcohol drinking (n=2), systemic eruption (n=1) and such as obtaining medications and supplies, glucose monitoring and overall cerebral infarction (n=1). glucose control. Patients were adult patients that spent at least 2 overnights SGLT2 inhibitors seemed useful in the treatment of Japanese patients in the hospital and sent home on any oral antidiabetic medication. with type 2 diabetes mellitus for 2 years. 70 patients (49 phone; 30 online) completed the survey. Results in Table. 2421‑PUB Table. Key Survey Results for Patients Discharged Home on Oral Agents and Noninsulin Injections. Effect of Multimodal Approach with Newer OAD’s on Glycemic Parameters and Insulin Dose in Indian Type 2 DM Patients Uncon‑ Medication Sulfonylurea Metformin TZD (6.3%) DPP-4I (13.9%) GLP-1 (5.1%) SGLT2 (3.8%) Insulin (17.7%) trolled on Insulin at Admission (34.2%) (87.3%) BALAJI JAGANMOHAN, SR., CHETAN PATIL, JR., ANJANA RAMAMURTHY, JR., (%) VAMSI KOLUKULA, SR., Bangalore, India, Mumbai, India, Hyderabad, India Medication at Sulfonylurea Metformin TZD (5.4%) DPP-4I (14.9%) GLP-1 (5.4%) SGLT2 (2.7%) Insulin (20.3%) Aim: To evaluate the effect of add on therapy with newer oral antidiabetic discharge (%) (36.5%) (81.1%) (OAD) agents in type 2 diabetes mellitus (T2DM) patients uncontrolled on Started on 35% % doing Home Before Hosp After DC (78%) insulin. Oral Agent Glucose (92%) Method: Total 20 T2DM patients uncontrolled on insulin and treated using inpatient Monitoring multimodal approach by addition of newer OADs were analysed retrospec- % Patient told 60% % not 26% (90% of tively. Mean daily dose (MDD) of antidiabetic drugs, HbA1c, fasting plasma to restart given any them restarted glucose (FPG) and postprandial plasma glucose (PPG) were analysed after 3 Prehosp instructions prehosp meds) months of multimodal approach therapy. Meds Results: Mean age of patients was 61.25±12.86 years, with 60% of male. % on DC with 77% % Problem 1% % on DC with 70% % Problem 4% Among these 80% of patients were suffering from diabetic peripheral neu- meds already getting HGM supplies getting Strips ropathy. Complications like, retinopathy, microalbuminuria, acanthosis and at home medication already at cataract were present in 35%, 20%, 15%, 5% patients respectively. MDD home of OADs at baseline and at 3 months are given in the Table. HbA1c, FPG % who 79% % who 21% % unable 3% and PPG reduced significantly by 1.83%, 51.75 mg/dl and 58.44mg/dl respec- Knew who to needed to to reach tively from their respective baseline of 9.4 ± 1.6%, 176.9 ± 65.24 mg/dl and contact with contact provider problem provider 236.4 ± 71.02 mg/dl (P- .0001, .0029, .0234 respectively). MDD of insulin was reduced significantly from 48.10 IU to 28.30 IU (mean difference: -19.8IU; Glucoses after First 2 days: First 2 Day First 2 days 2-14 days 2-14 days >300 p-.0001). DC to home <70 70-200 >300 <70 70-200 (mg/dl) Conclusion: Multimodal approach with newer OADs significantly improved glycaemic control in previously uncontrolled T2DM on insulin therapy with a never 84%; 33% 75-99% never 91% never 86%; 33% 75-99% never 91% remarkable reduction in daily insulin dose. of time; 56% of time; 56% everytime everytime

Summary: After discharge most patients on oral hypoglycemic agents already had their medications at home and restarted them. They also had glucose monitoring testing supplies and returned to testing. Glucoses were

ADA-Supported Research For author disclosure information, see page A751.

A631 CLINICAL THERAPEUTICS/NEWCATEGORY TECHNOLOGY—ORAL AGENTS

Table. Mean Daily Dose of Antidiabetic Drugs. 2424‑PUB Oral antidiabetic drug At Baseline At 3 months Sodium-Glucose Cotransporter-2 Inhibitors as Treatment for Diabe‑ tes Mellitus Type 2 in a Cohort of Patients Who Suffer Heart Failure N Mean daily dose N Mean daily dose JOSE IGNACIO MORGADO GARCIA DE POLAVIEJA, JESSICA ROA-GARRIDO, (mg) (mg) PILAR RODRIGUEZ ORTEGA, JOSE FRANCISCO DIAZ FERNANDEZ, Huelva, Spain Metformin 15 1603.3 19 1526.32 Introduction: Recently, it has been discovered a Sodium-Glucose Co- Glibenclamide 1 10 — — Transporter-2 Inhibitor (SGLT2-inhibitor) which has reduced cardiovascular Gliclazide 2 180 6 136.67 mortality and hospitalizations due to HF. We do not know best treatment of DM in these patients yet. Glimepiride 9 4.6 14 3.64 Purposes: To evaluate safety and efficacy of SGLT2 inhibitors in a cohort Pioglitazone 3 27.5 3 7.5 of patients with T2DM and HF. Saxagliptin 1 100 — — Methods: A Prospective, observational, study on T2DM patients with gly- Sitagliptin 1 100 4 100 cohemoglobin: A1c higher than 6.5% with oral antidiabetic medications and HF (admission in the last 6 months or high levels of N-terminal pro b-type Teneligliptin 2 20 2 20 natriuretic peptide (NT-proBNP) in addition to HF symptoms). SGLT2 inhibi- Vildagliptin 7 85.7 4 100 tors were started at stable situation (same treatment of HF in the last 3 Linagliptin — — 4 5 months). We assessed metabolic control, HF situation and safety (hypogly- Gemigliptin — — 4 50 caemia). Results: We analyzed 11 patients, 7 receiving canagliflozine, 3 empagliflo- Acarbose 1 50 — — zine and 1 dapagliflozine. Basal characteristics: Mean age 66.4 ± 8.6 years. Voglibose 3 0.5 — — Mean LVEF was 40.4±10%, mean NT-proBNP level: 1430±1243 pg/mL, mean Canagliflozin — — 8 100 of dose of furosemide was 52.7 ±38.2 mg daily. Mean BMI: 27.9 ± 4.2 kg/ m2 and A1c 8.45 ± 1.43%. Follow-up (120 days): A significant reduction was observed in NT-proBNP values (mean 912± 900 pg/mL, p=0.018), BMI (mean 2422‑PUB 26.2±3.9 kg/m2, p=0.001), A1c (mean 7.5±0.8%, p= 0.012) and dose of furose- Four-Year Outcomes of DPP-4 Inhibitors and Glinides Combination mide (mean 49.1 mg±37.3 mg daily, but p=0.34). Furthermore, it was reduced Therapeutics Therapy the rates of admission for HF (from 1.1±0.8 120 days before the treatment to Clinical Diabetes/ MASAYUKI HIRAMA, TOKIKO UEHARA, SHINYA KANEKO, YOSHIHIKO YAMADA, 0.01±0.3 120 days after, p=0.002). PUBLISHED ONLY Atami, Japan Conclusions: This pilot study shows the beneficial effects of SGLT2 inhibi- Object: Sulfonylurea (SU) use has been linked with increased risk of mac- tors in patients with Heart Failure, improving metabolic control and HF situ- rovascular disease. However, SUs are widely used. We focused on DPP-4 ation. This study has several limitations, such as small sample size and short inhibitors and glinides combination therapy as an alternative to SUs. DPP-4 follow-up. So, further studies, specially randomized trials, are required to inhibitors, do not usually cause hypoglycemia. Glinides improves postpran- contrast these hypotheses. dial hyperglycemia. DPP-4 inhibitors and glinide are seemed to be ideal combination for the purpose of avoiding macrovascular complications. We 2425‑PUB verify the efficacy and safety of switching to DPP-4 inhibitors and glinides A Clinical Case Series of Use of SGLT2 Inhibitors as a Add-On Ther‑ combination therapy from SUs. apy in Patient with T2DM and HIV Methods: This study was a Single center, retrospective, observational SOWMYA CHANDRA REDDY, KULSUM BANO, PAPITA ROZARIO, WILLIAM study that assessed the efficacy and safety of DPP-4 inhibitors and Glinide MOORE, AMY H. WARRINER, Birmingham, AL, Little Rock, AR combination therapy. Switchers who were switched to Vildagliptin and Background: People living with HIV (PLWH) frequently develop diabetes Glinides (Nateglinide or Mitiglinide) Combination therapy from SUs, were mellitus (DM), prediabetes, and metabolic syndrome. PLWH are believed to compared to “Continuers” who continued using SUs. Cohorts were matched be at high risk for metabolic disorders due to both traditional risk factors, by 1:1 propensity score matching to compare 4-year outcomes (18 patients such as age, sex, obesity, as well as risks of HIV infection and antiretro- in each cohort, mean HbA1c 8.35±1.38%, mean BMI 23.6±4.6, mean duration viral therapy. SGLT2 inhibitors have been associated with an improvement of diabetes 8.2±6.8 years in switchers and mean HbA1c8.2±1.57%, mean in HbA1C, diabetic nephropathy, blood pressure and weight loss in studies BMI 24.0±3.9, mean duration of diabetes 8.2±4.8 years in continuers). among HIV-negative persons. Here, we review a case series of patients with Results: At 4 years, the mean changes in HbA1c from baseline were DM and HIV who were initiated on SGLT2 inhibitors to evaluate the effec- -0.86% ± 0.9% vs. +0.1% ± 0.8% in Switchers vs. Continuers, respectively tiveness and safety of this class of drugs on glycemic control, weight, blood (P =0.02). Hypoglycemia risk was lower with Switchers (n=2) vs. Continuers pressure changes, side effect profile, and tolerance. (n=9). Body Weight decreases in Switchers (mean:-0.7kg), While increases Methods: PLWH with DM who were started on an SGLT2 inhibitor during in Continuers (mean: +2.3kg). routine clinical care, seen in a tertiary care HIV endocrinology clinic, were Conclusions: The study suggests that switching to DPP-4 inhibitors and selected for the case series (N=19) during a 2-year study period. Data on glinides combination therapy from SUs may improve glycemic control with- demographics, DM treatment, metabolic and biochemical variables were out increasing risk of hypoglycemia and undesirable weight gain. collected retrospectively. Results: Mean age of the patients was 49.7 years (SD 8) of which 58% 2423‑PUB were males and 42% females; 68% African American and 32% Caucasian. The mean duration of diabetes was 7.9 years (SD 6.2), mean baseline HbA1c WITHDRAWN 9.1% (SD 2.1). Mean HbA1c decreased 0.9% (SD 1.6), weight decreased by mean 3.4 kg (SD 6.6). A decrease in systolic blood pressure was seen in one- third of patients. Up to 50% of the patients developed a GU infection, pre- dominantly in women. Approximately half of the patients stopped the SGLT2 medication with reasons including acute kidney injury, polyuria, hyperkale- mia, and incontinence. There was no significant change in CD4 seen in any of the patients. Conclusion: Our study found that use of SGLT2 inhibitors appears safe and effective in PLWH. We found that there is significant decrease in HbA1c when used as add-on therapy. No unexpected adverse events were seen.

For author disclosure information, see page A751. ADA-Supported Research

A632 CLINICAL THERAPEUTICS/NEW TECHNOLOGY—PHARMACOLOGICCATEGORY TREATMENT OF COMPLICATIONS

CLINICAL THERAPEUTICS/NEW TECHNOLOGY— PHARMACOLOGIC TREATMENT OF COMPLICATIONS

2426‑PUB Clinical Dashboard to Evaluate and Monitor Institutional Quality and Effectiveness of Diabetes Crisis Management KALMAN E. HOLDY, JACQUELINE S. THOMPSON, ANGELIC M. MELLINA, DOROTHY WAGEMESTER, MARTY ENGLE, CAROLYN COX, San Diego, CA The algorithm for diabetes crisis (DKA/HHS) treatment is well established, but practices vary widely. Few studies systemically compare practices. We are systematically measuring DKA/HSS care and report our initial dashboard examining 2,700 admissions over 4 years. Table 1 lists the dashboard variables, updated quarterly. Cases are identified by diagnostic codes for DKA/HHS and/or use of specific order sets. Microsoft Excel and Tableau Software are used for analysis. Figure 1 displays selected items. The dashboard presents data to frontline caregivers to decrease variations and measure effectiveness of improving the application of the DKA/HHS algorithm. We are improving the dashboard by adding variables such as the time spent in the ER, the time to start insulin infusion, and the time transfer from the ICU. 2427‑PUB Determination and Evaluation of Hypertension in a Nonhuman Pri‑ mate Model of Type 2 Diabetes YINAN LIANG, ZUNYUAN YANG, ZHIGANG LIANG, YANQIN YU, YUANHAI CHEN, ZUNWEI YAO, YUBO SHEN, BARBARA C. HANSEN, LI GONG, WEN ZENG, Chengdu, China, Gulfport, FL

The common rat and canine hypertensive models do not fully reflect important Therapeutics

functional and structural features of hypertension in humans. We have estab- Clinical Diabetes/ lished NHP standards for normal blood pressure in a colony of 208 adult rhesus PUBLISHED ONLY monkeys and carried out a preliminary study of the efficacy of the combination anti-hypertensive drug, Entresto (sacubitril/valsartan) in rhesus monkeys with type 2 diabetes (T2DM). Blood pressure, fasting plasma glucose (FPG), HbA1c, total cholesterol (TC) and triglycerides (TG) were measured under ketamine anesthesia. To determine the normal range of blood pressure in adult rhesus monkeys, we excluded those meeting the following criteria: age <7 yrs or >15 yrs; SBP >140 mmHg or DBP >90 mmHg; body weight >14 kg; FPG >5.5mmol/L or HbA1c >4.5%; TG >1.1 mmol/L or TC >4.75 mmol/L. In the remaining 57 monkeys, considered to be healthy and normal, the normal blood pressure for NHPs was determined to be: SBP: 119±2 mmHg; DBP: 58±1 mmHg, Mean±SE. To compare efficacy in humans to rhesus, recently approved drug to treat heart failure, Entresto, also shown to be beneficial for the control of blood pressure, was used to evaluate efficacy in hypertension in the rhesus monkey model of mild T2DM. Six male monkeys (age: 14.2±1.0 yrs; FPG: 101.2±1.5 mg/dL; SBP: 140±2 mmHg; DBP: 67±3 mmHg) were randomly assigned to the Entresto group (N=3, p.o., 6.48/6.85mg/kg, q.d.) or to the vehicle group (N=3, vehicle p.o., q.d.). After 3 months of treatment, the blood pressure of the Entresto group was significantly lowered (SBP: -18.62%; 140±5 to 114±1 mmHg, p<0.05 and DBP: -18.30%; 68±6 to 56±1 mmHg), while the vehicle group were unchanged (SBP: 40±2 to 134±5 mmHg; DBP: 66±2 to 64±2 mmHg; p’s=NS). In conclusion, NHPs have similar normal blood pressure to humans, similarly develop spontaneous hypertension, and respond to this antihyperten- sive drug similarly to humans, thus providing an excellent animal model to validate the efficacy of anti-hypertensive drugs.

2428‑PUB Effect of Fenugreek Furostanolic Saponin on Cardiometabolic Parameters in Obese Subjects SREEJAYAN NAIR, DEREK SMITH, ANAND NAIR, RAMA NAIR, DEBASIS BAGCHI, ANAND SWAROOP, Laramie, WY, St. Louis, MO, Houston, TX, Piscataway, NJ We have recently reported that furostanolic saponins (FS) from the spice fenugreek, improves insulin sensitivity, reduces blood glucose and has a broad safety profile. In this study, we evaluated the effect of FS on cardiometabolic parameters in obese subjects. In this randomized, double- blind, placebo-controlled, study, 28 male or female subjects with body mass index >25 Kg/m2 and satisfied one of the following 3 criteria, HbA1c >5.6%, 2 h-post-challenge glucose > 125 mg/dL, HOMA-IR > 2.5, were randomly assigned to receive either placebo or FS (Fenfuro-TM, 500 mg bid) capsules for 12-weeks. Blood pressure, body weight, DEXA, glucose tolerance, serum lipids, and markers of inflammation and oxidative stress were assessed. HOMA-IR and serum 8-isoprostane were significantly lower in in subjects receiving FS. No significant changes were observed in any other parameters. Collectively, this pilot study suggests that fenugreek furostanolic saponins may improve insulin sensitivity and oxidative stress in overweight subjects. Supported By: National Institute of General Medical Sciences (1U54GM104944- 01A1 to S.N.)

ADA-Supported Research For author disclosure information, see page A751.

A633 HEALTH CARE CATEGORYDELIVERY—ECONOMICS

HEALTH CARE DELIVERY—ECONOMICS 2431‑PUB A Real-World Comparison of Outcomes in Patients Receiving Three 2429‑PUB Oral Antihyperglycemic Therapies, GLP-1, or Basal Insulin: An Effects of Pharmacist-Led Community Diabetic Care Using a Digital Analysis of Electronic Medical Record Data Solution in China: A Retrospective Analysis of Large, Real-World LAWRENCE BLONDE, ELISHEVA LEW, DENIS RACCAH, JULIANA MEYERS, Samples MAYANK AJMERA, KEITH DAVIS, MONICA BERTOLINI, BRUNO GUERCI, New YINGJIE LI, ZHEN WANG, YING CHEN, KAI LIU, Shanghai, China Orleans, LA, Paris, France, Marseille, France, Research Triangle Park, NC, Durham, Objective: To identify the effects of diabetes management using a digital NC, Vandoeuvre-les-Nancy, France solution by retail pharmacists for diabetic patients in a retrospective, obser- Objectives: Many treatment guidelines tend to recommend and clinicians vational analysis. usually follow a stepwise approach that may not be ideal as patients may Methods: We developed a digital healthcare solution that includes Blue- experience uncontrolled HbA1c between steps. Achieving HbA1c control tooth-enabled glucose meter, a mobile App, and a cloud database with 5 earlier and maintaining control longer is a key goal of T2DM treatment. To different algorithms for diabetic care support. This solution was commer- assess treatment in a real world setting, this study compared outcomes in cialized and deployed in 1,146 retail pharmacies of 165 brands in 94 cities patients with uncontrolled T2DM receiving 3 oral antihyperglycemic (OAD) in China to help pharmacists manage diabetic patients who were their therapies, GLP-1, or basal insulin (BI) with a retrospective electronic medical pharmacy members. During Jun 6th 2015 to Nov 4th 2016, a total of 85,790 record database. patients received in-store diabetic care and 3,653 of them were selected in Methods: Patients with a T2DM diagnosis between 1/1/2007 and this analysis following 4 inclusion criteria: 1) baseline fasting blood glucose 12/31/2014 were identified in the GE Centricity database. Patients receiving 3 OAD’s, GLP-1, or BI were selected (sequential or simultaneous initiation of (FBG) ≥ 7.0 mM or baseline random blood glucose (RBG) ≥ 7.8 mM, 2) the 3 OAD’s, GLP-1, or BI termed index date). Patients were required to have 6 intervention duration between the first- and last-time diabetic care was≥ 1 months pre- and 12 months post-index date physician history and a pre-index month, 3) the number of received diabetic care was ≥ 3 times, and 4) the baseline characteristics were complete. Among them, 1,657 received FBG date HbA1c >7%. HbA1c was compared between the 6 months pre- and 12 tests, 2,286 had RBG tests and 290 had both. months post-index date. Results: FBG and RBG were reduced on average by 0.8 mM (baseline: Results: A total of 59,598 patients met the inclusion criteria (23,450 3 9.1 ± 2.1 mM, last time: 8.3 ± 2.2 mM, n=1,657, P< 0.001) and 1.5 mM (base- OAD’s, 8,023 GLP-1, and 28,125 BI). The BI cohort had a higher baseline HbA1c line: 11.3 ± 3.4 mM, last time: 9.8 ± 3.4 mM, n= 2,286, P<0.001), respectively. (mean [SD, median] 9.7 [2.0, 9.2]) vs. 3 OAD’s (8.7 [1.5, 8.2]) or GLP-1 cohorts Therapeutics (8.7 [1.4, 8.3]) (P<0.0001). The BI cohort had the largest decrease in HbA1c

Clinical Diabetes/ Disease staging was improved with 482 patients of the FBG population during follow-up (i.e., decrease of 1.4% for BI indicating a 14.0% change, PUBLISHED ONLY (29.1%) and 709 patients of the RBG population (31.0%) returning to normal. The predominant glycemic control effect occurred after the first-time inter- 0.85/9.8% for 3 OAD’s, and 0.9/10.3% for GLP-1). Despite such marked vention, and the BG level remained stable throughout the follow-up. Simula- HbA1c decrease, 46.5% of patients in the BI cohort, 33.6% in the 3 OAD’s tion analysis revealed linear descending relationship of both FBG and RBG to cohort and 34.1% in the GLP-1 cohort had Hba1c > 8.0% during follow-up. time, and the concentration distributions were moving towards the normal Conclusions: In a large clinical practice database, despite improvements BG level. The linear fitting functions were: FBG, Y = -0.010x + 7.832; RBG, in HbA1c, over a third of patients had HbA1c > 8% during follow-up, advocat- Y = -0.018x + 9.574. ing the desirability of new treatment options or earlier more intensive use Conclusion: Digital healthcare solution is effective in assisting retail phar- of present therapies that could provide more robust and sustained glycemic macists to deliver community diabetic care in real world. control. Supported By: Sanofi-Aventis 2430‑PUB Impact of an Integrated Clinical Pharmacy Service on Hemoglobin 2432‑PUB A1c Reduction at a Federally Qualified Health Center Cost Effectiveness Analysis of a Flash Continuous Glucose Monitor‑ BENJAMIN CHAVEZ, EMILY KOSIROG, MARILYN BANH, Aurora, CO ing System for T1DM Patients Receiving Intensive Insulin Purpose/Background: Clinical pharmacy services (CPS) have been on the RICHARD HELLMUND, BIJU VARUGHESE, Alameda, CA rise lately with increasing data being published on their outcomes. However, Background: A novel, factory-calibrated flash continuous glucose monitor- ™ data on these types of services in low-income settings, such as federally ing system (FreeStyle Libre system) has features compared with other con- qualified health centers, is scant. This study aims to measure the impact of tinuous glucose monitoring (CGM) devices that may impact payer value. An one such service on patients with diabetes. Patients in this health center initial cost-effectiveness analysis has been performed to assess the value are referred for one-on-one CPS by their primary care provider to manage of flash CGM compared with specialized CGM for T1DM patients receiving specific disease states, with most those referrals being for the management intensive insulin. of diabetes. Pharmacists can initiate and modify medication therapy, order Methods: The IMS Core Diabetes Model was run over a 50-year horizon, laboratory tests, and recommend lifestyle changes per collaborative prac- modelling a population reflecting the IMPACT study cohort (Bolinder, 2016). tice agreement. Efficacy in terms of HbA1c and hypoglycemia was assumed to be the same Methods: Patients diagnosed with diabetes who had an initial visit with for both technologies due to a lack of reliable comparative data. Flash CGM the CPS team between July 1, 2015 and March 31, 2016 were included in this has a utility benefit (0.03; Matza, 2015) given the 91% reduction in self-mon- study. Individual charts were analyzed for improvements in hemoglobin A1C itoring of blood glucose (SMBG) in IMPACT. This benefit was not applied to (HbA1c), improvements in blood pressure (BP), number of visits, and specific specialized CGM since SMBG tests are required for calibration. CGM costs medication use. Data for these patients was collected through September were added to the SMBG annual costs in IMPACT. Flash sensor cash price was 59.90 (up to 14 days). CGM costs were 6124 (year 1) and 5400 30, 2016. € € € Results: A total of 250 patients were identified for review; 211 had a (later years). pre- and post-pharmacy intervention HbA1c. The average pre-intervention Results: Flash CGM was associated with 42% less direct cost and a 0.63 HbA1c was 10.7% +/- 2.5%; the average post-intervention HbA1c was 9.3% increase in QALYs. +/- 2.1% (p<0.0001). Patients whose pre-intervention HbA1C was > 9% had a Table. statistically greater decrease of 1.8% (p<0.0001) than patients with a base- Flash CGM Specialized CGM Increment line HbA1c of < 9%. A significant correlation was found between number of Direct costs $136,546 $233,530 -$96,983 visits and HbA1c difference. Statistically significant BP reductions were also found in patients who were hypertensive when they began CPS. LY 20.72 20.72 0.00 Conclusion/Implications: Patients with diabetes who had one-on-one QALY 12.93 12.30 0.63 visits with clinical pharmacists had significant improvement in their HbA1c and BP. Data suggests that patients with a HbA1c > 9% may benefit more Conclusions: This initial analysis shows flash CGM may be better value from this interprofessional consultation. This collaborative care model and than specialized CGM for T1DM patients receiving intensive insulin. More outcomes could be replicated at other institutions. work, including scenario testing, is needed to assess the value of these technologies. Supported By: Abbott Diabetes Care

For author disclosure information, see page A751. ADA-Supported Research

A634 HEALTH CARE CATEGORYDELIVERY—ECONOMICS

2433‑PUB 2435‑PUB Cost Calculation and Real-World Use of a Flash Continuous Glucose The Efficacy, Safety, and Cost Effectiveness Evaluation of Dipepti‑ Monitoring System for Patients Using Intensive Insulin: A UK NHS dyl Peptidase-4 Inhibitors in Patients with Type 2 Diabetes Perspective YUFEI FENG, TING LI, LEI XU, XUELIN SUN, XIN HU, Beijing, China, Shenyang, RICHARD HELLMUND, Alameda, CA China Background: A real-world dataset has been created from utilization of a Objective: To evaluate the efficacy, safety and cost-effectiveness of the novel, sensor-based flash continuous glucose monitoring system (FreeStyle five dipeptidyl peptidase-4 inhibitors (Alogliptin, Linagliptin, Saxagliptin, Libre™ system). When users scanned their sensor with a reader, they col- Sitagliptin, and Vildagliptin) in patients with type 2 diabetes mellitus. lected up to 8 hours of glucose readings that could be uploaded via internet- Methods: Search of Medline and Embase databases was conducted connected PC-based software to a de-identified database. A cost calculation to identify qualified studies in which DPP-4 inhibitors (DPP-4i) were used is presented from a UK NHS perspective based on the real-world scanning as either monotherapy or dual therapy (plus metformin). The primary out- frequency observed, comparing the acquisition cost of the flash monitoring comes were the mean change of HbA1c from baseline and the proportion of system with SMBG for patients receiving intensive insulin. patients achieving the goal of HbA1c <7%. The secondary outcomes included Methods: Unit costs were £0.04 per lancet and £0.29 per strip, and the mean change of body weight and the ratios of patients experiencing the sensor cash price was £48.29 per sensor. Sensor duration is up to 14 hypoglycemia. All the available data from the qualified studies were com- days. RCTs in T1DM and T2DM patients receiving intensive insulin showed pared via network meta-analysis. The binary or continuous models based on patients who used the flash monitoring system also used 0.3-0.5 SMBG Bayesian model with random effects were used. Cost-effectiveness analy- tests/day on average. sis was performed with the setting of pricing in China. Results: The real-world data showed the mean scan rate per day was 16 Results: No significant differences were found in the HbA1c reduction and per patient (median = 14), based on over 50,000 readers uploaded to the the proportion of patients achieved the goal of HbA1c <7% among five DPP-4 dataset. Annual cost of 16 SMBG/day is £1927.20 per patient compared with inhibitors, whether as monotherapy or combined with metformin. DPP-4i in £1255.54 per patient for flash monitoring sensors. The annual cost of addi- general do not significantly increase the risk of hypoglycemia when used as tional SMBG tests for patients using flash monitoring is £60.23, assuming monotherapy or dual therapy with metformin compared to placebo. Cost- 0.5 SMBG/day. The aggregate cost of glucose monitoring for flash monitor- effectiveness analyses showed that to reduce HbA1c by 1%, the annual cost ing system users is £611.43 (32%) less than for SMBG users. of Alogliptin is the lowest among DPP-4i when used as monotherapy or Conclusion: To achieve the rate of real-world glucose monitoring for combined with metformin. On the other hand, for every additional 1% HbA1C SMBG users that has been observed for flash monitoring system users reduction on top of metformin, Vildagliptin exhibits the lowest cost. Therapeutics would cost the UK NHS an additional £611.43 per patient per year. Com- Conclusions: All these five DPP-4 inhibitors show good efficacy, safety and Clinical Diabetes/ PUBLISHED ONLY pared to SMBG, flash monitoring is an affordable system to enable patients economic affordability. With the price setting in China, Alogliptin exhibits using intensive insulin to check their glucose frequently as recommended in the lowest cost for every 1% HbA1c reduction when used as monotherapy or clinical guidelines. combined with metformin while Vildagliptin shows the lowest cost for every Supported By: Abbott Diabetes Care additional 1% HbA1c reduction on top of metformin.

2434‑PUB 2436‑PUB Effectiveness of Team-Based Diabetes Care in Urban Cambodia Clinical Outcomes of Patients with Type 2 Diabetes Mellitus in a SOVANN PENG, PHANAVARINE MENH, NAOMI BYRNE-SOPER, MARIE E. Student-Run Clinic MCDONNELL, PETER CURRAN, MARGO HUDSON, Phnom Penh, Cambodia, NAVREEN K. PANDHER, FANGLONG DONG, KAY FANGEROW, AIRANI Boston, MA, Ketchum, ID SATHANANTHAN, Pomona, CA, Montclair, CA Background: Diabetes (DM) incidence in southeast Asia is expected to rise The purpose of this study is to assess clinical outcomes based on ADA by 70% in the next 20 years. In Cambodia (KH), over 200,000 people have Standards of Care in a student-run clinic for uninsured patients with type 2 diabetes and more than half are undiagnosed. Barriers to care include pov- diabetes mellitus (T2DM). A retrospective chart review was conducted for erty, high out-of-pocket costs, and lack of coordinated team-based chronic T2DM patients who sought care between 2014 and 2016. Data was collected disease management with trained specialists. We assessed outcomes of a at 3 time points: initial, a visit within the past 2 years, and the latest visit. new team-based DM program (TBDP) in urban KH. Among the 39 patients included, 64.1% were females, 94.9% Hispanics, and Methods: The HOPE Worldwide Medical Centers in Phnom Penh (PP) the mean age was 53.9 + 9.6 years. The average patient had diabetes for 7.1 started its first TBDP in 3/2013 funded by donations and patient co-pay- years. Comparing the most recent visit with the oldest visit, the mean weight ments. Personnel includes 4 MDs (internists), 2 RN educators, and 2 data decreased by 3.5 pounds (P=0.0297). Cholesterol, triglyceride, and LDL analysts. The model provides RN/MD co-visits, staff training programs, and decreased by 8.91 (p=0.0687), 12.05 (P=0.1724), and 6.42(P=0.3222) respec- support/group education for patients. tively. The average BMI decreased by 0.93 (p=0.0137). However, the average Results: A total of 1,111 new pts were enrolled in an electronic registry A1c increased from 7.45 (SD=1.40) to 7.77 (SD=1.63) and the HDL decreased from 7/2015-6/2016 and follow-up data were collected through 11/2016. Of by 2.57 (P=0.0565). At the most recent visit, 25.6% had an elevated blood those enrolled, 93% had T2DM, 1% T1DM and 6% pre-DM. 36% were previ- pressure. Over the last 2 years, each patient received 6.4 foot exams on ously undiagnosed. Mean age was 57.9 ± 10.7, BMI 24.2 ± 4.1, 67% female average. 11 of the patients were documented to have received a referral and 66% lived outside of PP. Mean baseline A1c (N=1091, 98%) was 9.0 ± for an eye exam since their initial visit. 51.3% (28.6% moderate intensity) 2.5. A1c was >7 in 75% and >9 in 43%. 7% of pts were taking insulin alone, were taking a statin, 26.3% were on aspirin, and 66.7% were on an ACEI at 3% insulin and orals, and 74% on oral meds only. While 891 (80%) pts had at their most recent visit. 8.3% of patients were documented to have received least 1 return visit, only 317 (29%) had a follow-up A1c. Over 12 months, the a flu vaccine at the clinic within the past year. Noncompliance with lifestyle overall mean absolute A1c change was -1.73 (P<0.001) and for pts with A1c changes and pharmacological treatment was reported by 48.7% and 76.9% >7%, -2.25 (p <0.0001). Of those with A1c >9 at baseline, 84% had a change respectively, with 41.2% reporting financial reasons for noncompliance. of -1% or greater. In conclusion, T2DM patients at a student-run clinic in Montclair, Califor- Conclusions: A TBDP was highly effective in lowering A1c in a low income, nia reached goals for medical history, physical exam, and lab evaluation with under-resourced Southeast Asian setting. Factors that may have influenced room for improvement in eye exams, immunizations, and self-management. outcomes include ability to attract and retain a large cohort, cost-sharing Status of eye exams and noncompliance for self-care could be explained by between the institution and patients, and selective utilization of follow-up a higher cost for specialty visits and medications. Overcoming financial cost A1c testing. Cost may be a factor in low follow-up testing rates and further may help improve these measures and enhance long term outcomes. research to address this is warranted. Supported By: PECO Foundation

ADA-Supported Research For author disclosure information, see page A751.

A635 HEALTH CARE CATEGORYDELIVERY—ECONOMICS

2437‑PUB 2439‑PUB Long-Term Impact of Early and Sustained A1c Control on Clini‑ Preoperative Optimization of Glycemic Control cal and Economic Outcomes: U.S. Veterans Health Administration CARLOS E. MENDEZ, NJERI WAINAINA, KURT PFEIFER, BARBARA SLAWSKI, (VHA) Database Analysis Milwaukee, WI FURAHA KARIBURYO, TAO FAN, LIN XIE, ADESUWA OGBOMO, ONUR BASER, It is well established that hyperglycemia the day of, during, and after PHILIP LEVIN, Ann Arbor, MI, Bridgewater, NJ, New York, NY, Towson, MD surgery predicts poor clinical outcomes. The section of Perioperative and Research has shown that effective A1c control in patients (pts) with T2D Consultative Medicine at the Medical College of Wisconsin has designed is associated with better clinical outcomes, while pts with poorly controlled an innovative initiative aimed to improve glycemic control in patients with A1c incur higher healthcare costs. This retrospective study evaluated the diabetes prior to elective surgeries. This comprises a nested preoperative impact of sustained A1c control and time-to-A1c control after T2D diag- diabetes clinic staffed by a physician diabetologist where patients with poor nosis on clinical and economic outcomes. The VHA database was used to glycemic control are scheduled for evaluation and intensification of diabetes identify pts diagnosed with T2D between Oct 1, 2006, and Sep 30, 2009. regimens. Appointments are made on short notice and patients are followed Index date was defined as first observed A1c record on or after Oct 1, 2006. closely by phone and in person until the surgery. Fructosamine levels are Participants needed to have ≥6 years of data: 1-year pre-index period (with used to assess short term glycemic control as needed. To evaluate the pro- no insulin prescription or >2 OADs); 2-year assessment period during which gram’s impact, records of initial 14 participants of the preoperative diabe- A1c sustainability was assessed; and ≥3-year follow-up period. Economic tes clinic were reviewed. Table 1 shows key characteristics of the studied and clinical outcomes were compared for pts with sustained A1c control vs. cohort. Mean A1c before the preoperative diabetes clinic visit was 9.5%, those with uncontrolled A1c (every A1c measurement <7.0% vs. ≥7.0% dur- reflecting poor glycemic control. After the intervention, mean fructosamine ing the assessment period), and for pts with early A1c control vs. late control levels obtained preoperatively were comparable to A1c levels of < 7%, dem- (A1c<7.0% during the whole assessment period vs. last 6 months only). Study onstrating significant improvement in glycemic control. Mean fasting glu- sample yielded 45,379 pts with early and sustained A1c control, 14,491 with cose the day of surgery was 138mg/dL with no hypoglycemia. No periopera- uncontrolled A1c, and 4,456 with late A1c control. Over the 3-year follow-up tive complications occurred in any patient. period, pts with sustained A1c control had lower total diabetes-related costs In conclusion, this innovative initiative resulted in significantly improved (−$3,877; P<0.0001) and lower total diabetes complications-related medical glycemic control preoperatively. Prospective studies are needed to confirm costs (−$1,393; P<0.0001) vs. pts with uncontrolled A1c. Early A1c control results. resulted in lower diabetes-related costs (−$3,385; P<0.0001) and lower total

Therapeutics Table 1. diabetes complications-related medical costs (−$674; P=0.0106) vs. pts with Clinical Diabetes/ late A1c control. Uncontrolled A1c was associated with a 17% higher risk of Sample characteristics (n=14) N or Mean (SD) PUBLISHED ONLY diabetes complications (95% CI=1.13-1.22; P<0.0001) and late control with a Age 57 (10.14) 14% higher risk (95% CI=1.07-1.21; P<0.0001). This real-world study suggests Sex (Female/Male) 12 / 2 that early and sustained A1c control in pts with T2D can have long-term Surgical Procedure (Ortho/other) 9 / 5 benefits including reduced risk of diabetes-related complications and lower diabetes-related costs. Pre visit A1c % 9.50 (2.22) Supported By: Sanofi U.S. Fructosamine µmol/L (n=9) 282.22 (52.61) Time to Surgery (days) 45 (32.72) 2438‑PUB Glucose day of surgery (mg/dL) 138 (42.69) Indirect Costs and Quality of Life for Patients with Type 2 Diabetes in Singapore KIAT MUN SERENA LOW, SU CHI LIM, XIAO ZHANG, JUANPING YANG, SU JIAN 2440‑PUB DARREN YEO, KUE LOONG KEVEN ANG, TAVINTHARAN SUBRAMANIAM, CHEE FANG SUM, Singapore, Singapore Despite growing prevalence of diabetes globally, information of its eco- WITHDRAWN nomic burden, particularly indirect cost and health-related quality of life (HRQoL), is scarce. We aim to study the indirect costs and HRQoL of patients with type 2 diabetes (T2D) in Singapore. This is a cross-sectional study using questionnaire survey administered on 203 patients with T2D who attended a Diabetes Centre in a hospital in Singapore in 2016. We used a human capital approach to calculate costs of absenteeism and presenteeism. Health utility scores and Visual Analog Scale (VAS) with 100 representing ‘best imaginable health state’ and 0 ‘worse imaginable health state’ were derived from EuroQol (EQ) 5D. Kruskall Wallis and Mann Whitney U-tests were performed to compare costs and health utility scores between groups. Spearman correlation was used to examine relationship between costs and health utility scores. The median workdays lost annually related to absenteeism and presenteeism were 4 days and 28 days respectively. The median annual costs of absenteeism and presenteeism were USD 16 and USD 146 respectively. The median EQ-5D index and EQ VAS scores were 0.83 and 70 respectively. The absenteeism costs tended to be lower in patients with lower socioeconomic status in terms of education (p<0.001), housing (p=0.007) and occupation (p<0.001). The presenteeism cost correlated moderately with EQ-5D index (r=-0.37; p<0.001). Patients who were single and not working tended to have lower EQ-5D index score than those who were married and working (p=0.02 and p=0.004 respectively). The EQ-VAS score was lower in those with poorer socio-economic status in terms of education (p=0.03), income (p=0.01) and occupation (p=0.016). Presenteeism contributed more substantially to indirect costs than absenteeism. The results provide useful baseline for comparing with future follow-up studies on patients with T2D and raise awareness of the impact of T2D on work productivity and HRQoL in Singapore. Supported By: Alexandra Health (STAR/16102)

For author disclosure information, see page A751. ADA-Supported Research

A636 HEALTH CARE CATEGORYDELIVERY—ECONOMICS

2441‑PUB 2443‑PUB Validation of an Insulin Infusion Nomogram for the Coronary ICU Gaps in Attendance in Young Adults with Type 1 Diabetes: Need for SEANN SETO, YVONNE KWAN, AMITA WOODS, Toronto, ON, Canada Quality Improvement Background: Hyperglycemia is associated with increased morbidity and JENNY YUJING WANG, XINYE SERENA WANG, PATRIZIA DIRAIMO, CHERYL mortality in acutely ill patients, including those treated in intensive care set- HARRIS-TAYLOR, LEAH J. DRAZEK, JANIS RUSEN, NICOLA A. FARNELL, KIS- tings. However, there is evidence showing that intensive glycemic control HANI SARVANANTHAN, VICTOR CIPRIANO, LORRAINE L. LIPSCOMBE, GEETHA is also associated with increased mortality as well as a higher incidence MUKERJI, Toronto, ON, Canada of severe hypoglycemia. Numerous studies have demonstrated that a stan- Background: The transition from pediatric to adult type 1 diabetes (T1DM) dardized insulin infusion nomogram improves the proportion of blood glucose care is associated with decreased medical follow-up, lost opportunities for levels within target range while reducing the risk of severe hypoglycemia. counselling, and increased risk of diabetes-related hospitalizations. However, the majority of these trials studied medical-surgical intensive care Objective: We aimed to characterize attendance patterns and reduce unit (MSICU) patients and data on coronary ICU (CICU) patients is lacking. non-attendance (no-show and cancellation within 24 hours) rates using the Objective: To compare the efficacy and safety of standardized insulin Model for Improvement quality improvement (QI) framework at a dedicated infusion nomograms vs. non-standardized intravenous (IV) sliding scales in clinic for young adults (YA) with T1DM at the University of Toronto-affiliated patients admitted to the coronary ICU. Women’s College Hospital (WCH) in Toronto, Canada. Methods: A single-centre, retrospective before-after cohort study com- Method: Non-attendance rates between February 2015 and September pared glycemic control in patients admitted to the CICU. Patients received 2016 were audited and plotted monthly on a Statistical Process Control regular IV insulin at rates specified by either non-standardized sliding scale (SPC) chart. Consecutive new patient charts from this period were reviewed (prior to May 2015) or standardized nomogram (May 2015 onward) titrated to for demographic data, metabolic control, and counselling rates. A patient a target range of 6-10 mmol/L (108-180 mg/dL). survey was administered between October 2016 and January 2017 to iden- Results: A total of 58 patients were enrolled into this pilot study. Patients tify root causes of non-attendance. Continuous variables were reported as in the nomogram group achieved a greater proportion of blood glucose levels median (interquartile range) and categorical variables in percentages. within target range compared to patients in the IV sliding scale group (42.4% Results: There were 450 scheduled visits for 150 patients during this vs. 28.3%; P<0.001). The nomogram group also achieved the target blood period; the mean non-attendance rate on SPC chart was 32% with no spe- glucose range more rapidly (15.4 hours vs. 25.8 hours; P=0.03). Hypoglycemia cial cause variation. Chart review [n=30, 84% female, age 18.7 (18.2, 19.7), (0.15% vs. 0.86%) and severe hypoglycemia (0% vs. 0.10%) occurred less duration of T1DM 8 years (6, 14), and A1c 8.8% (8.0, 10.1)] showed 9.7% of frequently in the nomogram group. referred patients were never seen, 18% were lost to follow-up, 39% were Therapeutics Conclusions: In patients admitted to the CICU requiring IV insulin, a stan- seen <3 times a year, and 74% had 1 non-attendance. Documented coun- Clinical Diabetes/

≥ PUBLISHED ONLY dardized insulin nomogram protocol resulted in a greater proportion of blood selling rates for hypoglycemia, pre-conception, and sick day management glucose levels being within the target range, while also decreasing the inci- were 68%, 68%, and 14%, respectively. In preliminary survey data (n=12), dence of hypoglycemia. patients highlighted fear of judgment as cause for non-attendance and preferred email reminders for appointments. 2442‑PUB Conclusions: Non-attendance is a significant issue in the YA T1DM clinic at WCH. Using iterative Plan-Do-Study-Act cycles, a reminder system and WITHDRAWN patient-centered conversation guide will be tailored to address root causes of non-attendance and outcomes will be evaluated.

2444‑PUB The Diabetes and Dementia (DIA-DEM) Project: Targeting the For‑ gotten Population AMAR PUTTANNA, KATELYN AITCHISON, PARIJAT DE, Birmingham, United Kingdom The combined co-morbidity of diabetes and dementia is on the rise due to both an increase in the ageing population and prevalence of diabetes. The DIA-DEM project is aimed at assessing current trends in hospital admissions, diabetes medication usage and need, complications and length of stay with an overarching aim to reduce health and economic burden by optimis- ing care and minimising variation. We analysed a retrospective cohort of patients coded with ‘diabetes’ and ‘dementia’ admitted to a large district general hospital between June and August 2016. One hundred and forty one patients (73 male, 68 female, mean age 82.5 yrs +/- 4.2 SD) with type 2 diabetes were included. Mean length of stay was 8.5 days (+/- 3.5 SD). Seven- teen (12.1%) had vascular dementia, 44 (31.2%) Alzheimer’s, 78 (55.3%) not coded. Fifty six (39.7%) were from a care home, 85 (60.3%) from their own home. Mean HbA1c was 55.6 mmol/mol (+/- 12.7 SD). Twenty five (17.7%) were on sulfonylureas, 28 (19.9%) on DPP-4 inhibitors, 63 (44.7%) were on metformin, 33 (23.4%) were on insulin, 25 (27.7%) were not on any medication, 6 (4.3%) were on sulfonylurea and insulin and 11 (7.8%) were on 3 or more agents. Twenty six (18.4%) had at least one episode of hypoglycaemia, clinically significant (<3 mmol/l) in 15 (10.6%). Twenty (14.2%) patients died within 6 months, associated with HbA1c>55mmol/mol. Applying our DIA- DEM principles, 74 patients (52.5%) would have had focused diabetes intervention with medication adjustment (stopping SU or insulin where needed), timely HbA1c review that would allow appropriate management of glycaemic abnormalities and help reduce length of hospital stay. This assessment of our DIA-DEM principles prove that much can be done in targeting this population, particularly those on high doses of sulfonylurea and complex insulin regimes that increase the risk of hypoglycaemia and prolonged length of hospital stay. More data and targeted approaches are required to enhance care of this population and is part of the ongoing DIA-DEM project.

ADA-Supported Research For author disclosure information, see page A751.

A637 HEALTH CARE CATEGORYDELIVERY—ECONOMICS

2445‑PUB 2447‑PUB The Value Index for SGLT2 Medications in Medicare Patients The Value-Index Analysis of Metformin Preparations in Medicare SAAD SAKKAL, Mason, OH Patients Could Save $100 Million Introduction: In the present paradigm, the yearly national expense for SAAD SAKKAL, Mason, OH diabetes is $340 billion while less the 14% of people with Diabetes achieve Introduction: In the present paradigm, the yearly national expense for dia- all targets together. Medicare has an estimated 13520473 beneficiaries betes is $340 billion while less the 14% of people with diabetes achieve all who have diabetes, with estimated annual costs of around $10 Billion targets together. A major part of the drop in adherence is medication pricing ($9554829391). We asked in this abstract: What is the value index for pres- and side effects. We asked in this abstract: What is the value index for present oral SLGT2 medications in Medicare patients? ent oral Metformins in Medicare patients? Design: We compared the value index by finding how much money ($) is Design: We compared the value index by finding how much money ($) is spent on improving HbA1c by 1% in three months. This calculation takes spent on improving HgA1c by 1% in three months. This calculation takes quality factor and cost factor for an estimate of the value of each medica- quality factor and cost factor for an estimate of the value of each medication. We used the price Medicare pays for each from the Medicare spending tion. We used the price Medicare pays from the Medicare spending data data for 2015. We assumed SGLT2 will drop HbA1c by % 1.03 which is the for 2015. From the literature, we assumed %0.6 HgA1c drop for generic and best available in the literature. %0,8 drop for non-generic. Results: SGLT2: 1) annual costs per patient were (Invokana: $2097, Farxiga: Results: 1) Cost per unit (tablet or capsule) for generic metformin was: 1739, Jardiance: 1467, and 1226 for Glyxambi, the latter is a combination) $0.07, for Metformin XR: $0.37, Glucophage: $1.37, for Glucophage XR: 1.05, for an average annual cost per patient of $1998. 2) The total patients were Fortamet: $28.40, Glumetza: $39.1. 2) The annual Medicare expense per user 245458 and annual spending was $490595264. 3) Diabetes care value index: was: generic $36.87, Metformin ER: $142, Glucophage: $747.8, Glucophage was calculated at $508 for Invokana, $422 for Farxiga, $356 for Jardiance, XR: $665. Fortamet: $12193, Glumetza $13915. 3) The diabetes care value and $306 for Glyxambi. 4) Comparing The relative expense to Jardiance was index was: $20 per 1% HgA1c drop for generic, $55 for Metformin ER, $233 0.85 for Glyxambi, 1.18 for Farxiga and 1.42 for Invokana. 5) The observed for Glucophage, $210 for Glucophage XR, $3810 for Fortamet, $4350 for Glu- direct annual price difference between the highest and the lowest priced metza. 4) From the perspective of value index, it appears there are three medication was: $202 per patient, but, the actual effective difference based tiers: most economical (metformin and metformin ER), intermediate: 4-10 on the relative value index is $286. 6) If the annual cost difference of $286 folds higher price (Glucophage and Glucophage XR) and the highest priced: per user is projected to all patients using the highest expense preparation 200 folds (Fortamet and Glumetza). 5) If the annual cost difference of $13000 Therapeutics (225029), the savings adds up to $64358294 or %13. 7) From the perspective per user (between Glumetza and Metformin) is projected to patients using Clinical Diabetes/ of value index, it appears there are Two tiers: First is more clinically useful the highest expense tier metformin preparations (10994), the savings could PUBLISHED ONLY and economically effective (Jardiance/Glyxambi) and the second is higher add up to $142922574 per year. priced (Invokana, Farxiga). Conclusion: the value index for using metformin preparations in Medicare Conclusion: The effective value index saving of $286 favors the most ($ spent to bring HgA1c by %1) still favors older agents over newer agents, economical preparation. A projection for the 225029 patients suggests a with an annual difference of $13000. A projection for the 10994 patients potential savings of $64358294. using high expense preparations shows a potential savings of $142922574 per year. 2446‑PUB The Value Index for DPP-4 Medications in Medicare Patients 2448‑PUB SAAD SAKKAL, Mason, OH Introduction: In the present paradigm, the yearly national expense for dia- WITHDRAWN betes is $340 billion while less the 14% of people with Diabetes achieve all targets together. Medicare has an estimated 13520473 beneficiaries who have diabetes, with estimated annual costs of $10 Billion ($9554829391). We asked in this abstract: What is the value index for present oral DPP-4 medications in Medicare patients? Hypothesis and Design: We compared the value index by finding how much money ($) is spent on improving HgA1c by 1% in three months. This calculation takes quality factor and cost factor for an estimate of the value of each medication. We used the price Medicare pays for each preparation from the Medicare spending data for 2015. We assumed DPP-4 will drop HgA1b by %0.66 which is the best available in the literature. Results: DPP-4: 1) annual cost per patient were (Januvia: $2572, Ong- lyza: 2304, Tradjenta: 2105, Nesina: 1769) for an average annual cost per patient of $2447. 2) The total patients were 1197066 and annual spending was $2929750365. 3) Diabetes care value index: was calculated at $974 per 1% drop of HgA1c in three month for Januvia, $872 for Onglyza, $794 for Tradjenta, and $670 for Nesina. 4) Comparing the relative expense to Nesina was: 1.18 for Tradjenta, 1.3 for Onglyza and 1.45 for Januvia. 5) The observed direct annual price difference between the highest and the lowest priced medication was: $803 per patient, but, the actual effective difference based on the relative value index is $972. 6) If the annual cost difference of $972 per user is projected to patients using the highest expense DPP-4 preparations (943269), the savings adds to $916957468 or % 31. 7) From the perspective of value index, it appears there are Two tiers: First is most economical (Nesina), and the second is highest priced (Januvia, Onglyza). Conclusion: The effective value index saving of $972 favors the most economical DPP-4 preparation. A projection for the 943269 patients suggests a potential savings of around $1 Billion per year ($916957468).

For author disclosure information, see page A751. ADA-Supported Research

A638 PEDIATRICS—OBESITYCATEGORY AND TYPE 2 DIABETES

2449‑PUB 2451‑PUB “Talking Was the Most Important Thing” Association between Iron Markers and Diabetes Risk in Indigenous JENNEY LEE, MARIT L. BOVBJERG, ROSA WOLFF, MICHAEL A. MAY, Corvallis, OR Children Living at High Altitude Complex chronic diseases burden the healthcare system. Telehealth VALERIA HIRSCHLER, CLAUDIO GONZALEZ, CLAUDIA MOLINARI, GUSTAVO solutions might reduce that burden and improve health outcomes. There is MACCALLINI, LUIS CASTANO, Buenos Aires, Argentina, Bilbao, Spain concern that patients may be slow to adopt “cold technologies” in place Previous studies showed that iron overload induces the formation of free of in-person interaction with healthcare providers. However, research on radicals and have deleterious effects on various cellular structures. We have patients’ experiences with such technologies are sparse, leaving open the previously found that Argentinean Indigenous children living at high altitude question of how telehealth platforms are perceived by healthcare consum- had higher haemoglobin levels than children living at sea level. The objective ers, or how readily they might be adopted by chronic disease patients. was to determine the association of four different markers of iron metabo- This pilot study for Medicaid patients with diabetes employed a tablet- lism such as serum iron, ferritin, transferrin, and transferrin saturation with based health monitoring application. Patients used a cellular-enabled glu- risk of type 2 diabetes in Indigenous children living at high altitude. cometer to upload blood glucose readings, which were monitored remotely In a cross-sectional study, 352 school Indigenous children (166 males) liv- by nurse case managers. Patients and nurses interacted via a text mes- ing at 3750 m above sea level with a mean age of 9.64+2.33 years were saging feature and by phone. We conducted in-depth, semi-structured exit examined in November 2012. Anthropometry, blood pressure, lipids, and interviews (n=27) to assess patient experiences. glucose levels were measured. The prevalence of overweight was 7.7% We found that, despite a relative lack of technological knowledge, (27) and obesity was 2.8% (10). Mean haemoglobin levels were 15.5g/dL. patients held an overwhelmingly positive view of the intervention. The There were no significant differences in age, BMI, waist circumference, iron feature viewed most positively by the patients was the ability to send and markers, and apo A and apo B levels between genders. Transferrin levels receive text messages with the nurses. The tablet became an extension of were positively associated with BMI (r0.24), glucose (r0.13), and apo B levels the provider’s office, facilitating patients’ connection to healthcare profes- (r0.27); iron levels were inversely correlated with BMI (-0.13) and systolic sionals. As one patient commented, “It was nice having that security there, blood pressure (r-0.13); and transferrin saturation was inversely associated that comfort of someone there caring about you.” The technology was not with BMI (r-0.13), systolic blood pressure (r-0.15), TG/HDL-C (r-0.12), apo B alienating; rather, the tablet became an important conduit for the nurse- (r-0.15), and HOMA-IR (r-0.14). Multiple linear regression analyses showed patient relationship. Patients experienced the tablet-based application not that transferrin levels were positively associated with age, BMI, and apo B as cold, but as imbued with emotionality. The perception of the tablet as a levels adjusted for confounding variables (R2=0.15). Furthermore, transferrin “warm technology” was key to patient satisfaction in this pilot project, and saturation was inversely associated with BMI and apo B (R2=0.10). Therapeutics may prove a useful concept in adoption of future telehealth interventions. High transferrin and low transferrin saturation were associated with dia- Clinical Diabetes/ PUBLISHED ONLY betes risk in Indigenous children living at high altitude with high iron levels. Our findings highlight the public health importance of monitoring iron mark- PEDIATRICS—OBESITY AND TYPE 2 DIABETES ers and also suggest that iron metabolism may contribute to future type 2 diabetes in this community. 2450‑PUB “I Just Forgot”: Barriers and Strategies for Glucose-Lowering Oral 2452‑PUB Medication Use and Impact on Adherence in the Treatment Options Higher Percent Body Fat in Infants of Diabetic Mothers vs. Healthy for Type 2 Diabetes in Adolescents and Youth (TODAY) Trial Term Infants in a Largely Hispanic Population ELIZABETH M. VENDITTI, PATRICE YASUDA, NANCY CHANG, LORI M. LAFFEL, ELIA N. ESCANAME, RACHEL L. JACOB, CRISTINA PENON, DIANA ANZUETO, GERALDINE H. MCGINLEY, NELLY MIRANDA, JEANIE B. TRYGGESTAD, NATALIE NICHOLAS ALANA, NICHOLAS CARR, JONATHAN KING, CYNTHIA BLANCO, San W. ABRAMSON, KENNY TAN, LINDA DELAHANTY, Pittsburgh, PA, Los Angeles, Antonio, TX CA, Boston, MA, Philadelphia, PA, Houston, TX, Oklahoma City, OK, Aurora, CO, Background: Limited data on body composition in infants of diabetic Rockville, MD mothers exists. Body composition early in life likely impacts later metabolic Youth with type 2 diabetes require lifelong medication use. Understand- alterations. ing individual adherence barriers and strategies for taking medications may Objective: Compare percent body fat (BF) in infants of diabetic mothers vs. aid in developing effective self-management interventions and preventing healthy term infants, in a largely Hispanic population using anthropometric, comorbidities. The aims of this analysis were to identify types and frequen- air displacement plethysmography (Peapod) and dual-energy x-ray absorpti- cies of barriers and strategies and their associations with baseline char- ometry measures. acteristics, and to explore whether strategies had an impact on adherence Design/Methods: Prospective longitudinal cohort of term singleton over time. Data from a subgroup of TODAY participants who maintained infants born to diabetic mothers control infants vs. control infants were adequate glycemic control on any of the randomized study treatments [met- assessed at birth and 3 months of age for anthropometric and skin fold formin (M) alone, metformin + rosiglitazone (M+R), metformin + lifestyle measurements, air displacement plethysmography and DXA. Enrollment eli- (M+L)] at Months 6 (N=566), 12 (N=478) and 24 (N=376) following randomiza- gibility required no major congenital anomalies, no postnatal complications, tion were analyzed. The cohort was 64% female, 79% racial/ethnic minori- AGA and no maternal chronic disease history in control infants. Statistical ties, mean age 14 years, <2 years diabetes duration, ≥85th percentile BMI, analyses were performed utilizing SPSS 22.0. and had adult caregivers who agreed to provide support. Study clinicians Results: To date, assessments of 38 healthy term infants born to moth- documented the presence of medication barriers (e.g., disruptions to daily ers with and without diabetes mellitus had a mean percent BF of 13.5% ± routines and other reported concerns) and strategies (e.g., rely on family, 5.06% vs. 9.92% ± 3.95% vs. and fat mass 470 g ± 230 g vs. 310 g ± 139 g schedule, reminder device) to improve adherence. At each time point, suc- respectively that was significantly different (p=0.016, p=0.008) between the cessful medication adherence (≥ 80%) was inversely related to the number groups. Percent BF remained significantly higher after adjusting for pre-preg- of barriers reported (p’s=.0001). “Forgetting” with no specific reason named, nancy BMI and maternal age by logistic regression (OR=1.23, CI=1.01-1.51, or disruptions in regular meals, sleep or schedule factors, accounted for p=0.03). Positive correlations were found between fat mass, arm circumfer- ≥ 60% of responses. No significant baseline associations with barriers or ence and thigh skin fold (R= 0.6, p<0.001) utilizing either Peapod or DXA. adherence strategies were found. Family support was the strategy identified Conclusions: Percent BF and fat body mass between term infants born to as helpful by most youth (≥50%), followed by the pairing of medications with mothers with and without diabetes mellitus immediately after birth were daily routines (>40%); the latter was associated with higher adherence rates higher in this cohort of primarily Hispanic mothers. Correlations between (p=0.009). These findings are similar to those reported in youth with other commonly used anthropometric measurements to estimate percent body fat serious chronic diseases. Prospective studies to enhance family support and are limited. Long term follow-up for body composition analysis with is ongo- self-management are warranted. ing. Methylation of key insulin signaling genes on cord blood samples from Supported By: National Institute of Diabetes and Digestive and Kidney Dis- both groups are pending. eases/National Institutes of Health (U01DK61212, U01DK61230, U01DK61239, Supported By: San Antonio Medical Foundation U01DK61242, U01DK61254)

ADA-Supported Research For author disclosure information, see page A751.

A639 PEDIATRICS—TYPECATEGORY 1 DIABETES

PEDIATRICS—TYPE 1 DIABETES 2455‑PUB The Development and Prototype Testing of a Patient-Centered 2453‑PUB mHealth App for Diabetes Experiences of Patients with Type 1 Diabetes during Transitioning BREE E. HOLTZ, KATHARINE M. MURRAY, DENISE S. HERSHEY, JULIE K. DUN- to Adult Care NEBACK, SHELIA R. COTTEN, AMANDA J. HOLMSTROM, ARPITA VYAS, MOLLY SUSAN PENA-ALMAZAN, ADRIENNE STOLFI, PAUL BREYER, Dayton, OH M. KAISER, MICHAEL A. WOOD, East Lansing, MI, Lansing, MI, Lubbock, TX, Ann Transitioning to adult care by pediatric patients with type 1 diabetes (T1D) Arbor, MI is a challenging task and there is limited information about the experiences Background: Type 1 diabetes (T1D) afflicts more than half a million chil- of these patients during the transition process. This study aimed to obtain dren. T1D is complex and parents often manage their child’s disease. Chil- information from patients with T1D at Dayton Children’s Hospital Diabetes dren must eventually transition to self-care, which is often difficult. Adher- Clinic who transitioned to adult care. ence to the prescribed treatment regimen drops during this time. Questionnaires were mailed to 102 patients who transitioned to adult Mobile health (mHealth) apps have been shown to be successful at moni- care, with 33 (32%) patients responding. Mean (SD) age at diagnosis was 9.8 toring and managing diabetes. Our proposed app, MyT1DHero is unique in (4.1) y, and at transfer was 19.2 (1.1) y. Mean A1c prior to transfer, obtained that it links the child’s information to their parent’s cell phone and promotes from clinic charts, was 8.6% compared to self-reported mean A1c of 7.6% positive communication within families. during the study. 39.4% of respondents were seen by an adult provider within Objective: The purpose of this study was to a use patient-centered 3 mo and 45.5% within 3-6 mo after transfer. Reasons for transfer included approach to inform the development of our app, MyT1DHero. We also con- reaching 22 years old (n=11), leaving for college or work (n=9), feeling too ducted prototype testing of the app with our target population to measure old for children’s hospital (n=3), more convenient location for work (n=1), and usability and satisfaction. pregnancy (n=1). Choice of current caregiver was dictated by referral from Methods/Development Process: Conceptualization of the app included the pediatric provider, referral from relative or friend, convenient location brainstorming sessions with 4 clinicians specializing in T1D and communi- for work or home, and the type of insurance provider. Adult endocrinologists cation experts. We conducted qualitative focus groups with kids with T1D followed 81.8% of respondents and 12.1% were followed as well by mental and their parents to learn about their management plan and communication health providers. Twenty-five (75.8%) respondents recommended 18-22 y as patterns, what apps they use, and feedback on our app. Results drove the the ideal age of transfer to adult care. 75.8% of respondents agreed that design and development of the app, created by a professional app devel- their pediatric provider was helpful in the transition process. Reasons cited opment company. Prototype testing was conducted in winter 2016. We Therapeutics recruited 10 parent/child pairs (n=20) to use the app for 4 weeks. Interviews

Clinical Diabetes/ included providing a list and information on adult endocrinologists, sending measured satisfaction of app usage. Server information was analyzed to PUBLISHED ONLY prompt referral and medical information to adult provider, and willingness to assist until transfer occurs. Respondents also gave suggestions to help ease quantify learning time, efficiency of use, engagement and user errors. transfer to adult care. Results: Prototype testing participants were satisfied with the app. There The experiences and impressions of the T1D patients in our clinic who were a few minor issues that were addressed. These findings informed app transitioned to adult care gave valuable insight on the transition process. refinement for use in the full intervention. Recruitment for that study is We have incorporated this information in the proposed transition program underway now. for our clinic and we will determine if these will help ease transitioning to Conclusions and Future Implications: Gathering insight from clinicians and adult care. patients about daily management of T1D supports the development of this app to aid in the transition to self-management. The results of prototype testing show promising impacts that we hope to see in intervention testing. 2454‑PUB Supported By: American Diabetes Association (1-16-ICTS-045 to B.E.H.) Games for Children with Type 1 Diabetes: A Systematic Review BREE E. HOLTZ, KATHARINE M. MURRAY, MOLLY M. KAISER, East Lansing, MI Background: More than half a million children under the age of 20 have 2456‑PUB type 1 diabetes (T1D). To ease the difficulties of management, video games Correlation of Proinsulin to C-Peptide Ratio with BMI and Age in have been tested to teach health skills, increase self-efficacy and improve Children with Type 1 Diabetes adherence. The popularity of video games has led to the creation of seri- DANIEL S. HSIA, BENJAMIN R. ELLISON, JEFFREY H. BURTON, Baton Rouge, LA, ous games to ease management of chronic diseases. The purpose of serious Waco, TX games is to focus on health management skills in an engaging and motivat- The proinsulin to C-peptide (PI:C) ratio has been proposed as a marker of ing way. β-cell dysfunction. Elevations in the PI:C ratio are seen in people at risk for Objective: The goal of this study was to review existing literature to deter- and diagnosed with type 1 and type 2 diabetes while reductions in the PI:C mine the impacts of serious games for children with T1D. ratio have been associated with improved β-cell function. Nine participants 2 Methods: The search resulted in 4,929 articles. Of those articles, 157 with a mean age of 13 years, a mean BMI of 21 kg/m , and within 5 years were focused on diabetes. Studies were further trimmed based on their use of type 1 diabetes diagnosis were enrolled in a single visit cross sectional of a game intervention, resulting in 7 articles that were assessed for this study. Their mean HbA1c was 8.5% (69 mmol/mol). Ten participants with a 2 review. mean age of 13 years and a mean BMI of 25 kg/m were enrolled as a control The studies included a total of 352 participants with a mean of 50.29 group. PI and C levels were measured concurrently after a 10 hour fast with (SD=47.29). Participant ages ranged from 3-38. Study duration varied from a levels below the assay lower limit of detection assigned a value of one-half 30-minute session in one study to 40 weeks in another. Only 3 were random- the lower limit of detection. Molar ratios of PI and C were calculated and ized controlled trials. multiplied by 100 to obtain PI:C ratios. PI:C ratios were significantly differ- Results: Four studies were usability tests, and 3 were randomized con- ent between the type 1 diabetes and control groups (6.56 vs. 1.46, p< 0.01). 2 trolled trials. Study duration was varied, and 4 studies had a strict regime A negative correlation was seen between the PI:C ratio and BMI (r = 0.63; 2 of play. Three studies found high comprehension, satisfaction, and enjoy- p= 0.01) as well as between the PI:C ratio and age (r = 0.45; p= 0.05) in the ment of the game. One found several usability and comprehension issues. type 1 diabetes group (Figure). These relationships were not observed in the The results were mixed for health and psychosocial variable outcomes. No control group. These results indicate that children with type 1 diabetes and significant health outcomes were found, and psychosocial variables were a higher BMI may have less β-cell dysfunction (lower PI:C ratio) and that chil- impactful in only 2 studies. No conclusion can be drawn between study dren diagnosed at a younger age may have more β-cell dysfunction (higher duration and significant findings. Strict regime of play did not appear to PI:C ratio) compared to older children. impact significant outcomes. Conclusion: This study reports the results of 7 studies that tested impacts of a video game on children with T1D. Unlike previous systematic review findings, positive outcomes in satisfaction, health or psychosocial variables were not found to be more frequent among studies that had a longer duration. This review provides important implications for future development of games for the management of T1D.

For author disclosure information, see page A751. ADA-Supported Research

A640 PEDIATRICS—TYPECATEGORY 1 DIABETES

Figure. 2458‑PUB Effect of Linagliptin Added to Insulin Treatment on Metabolic Con‑ trol, Insulin Dose, and Hypoglycemia in Patients with Uncontrolled Type 1 Diabetes MARIA L. MONTES DE OCA LOYOLA, CLAUDIA M. HERNÁNDEZ-ROBLES, EDGAR DURAN PEREZ, ALBERTO AGUILAR GARCÍA, MARÍA L. EVIA-VIZCARRA, EDEL R. RODEA-MONTERO, DIANA FARFÁN VÁZQUEZ, ANA K. VILLA-MARTÍNEZ, GEOR- GINA A. BARAJAS-MEDINA, FABIOLA ANGULO-ROMERO, MARÍA DE LOURDES REYES-ESCOGIDO, RODOLFO GUARDADO MENDOZA, León, Mexico Introduction: T1DM is characterized by autoimmune destruction of the pancreatic beta cells, and treatment is based on insulin. New studies have shown abnormalities in glucagon and incretin effect in T1DM. The goal of this study was to evaluate the effect of Linagliptin + insulin on metabolic control, insulin dose and hypoglycemia in uncontrolled type 1 diabetic patients. Methods: This was single group study, including adult patients with uncontrolled type 1 diabetes (HbA1c higher than 7%). The protocol was approved by the ethical committee. Patients were recruited from the Endocrinology Department, and they had a basal evaluation with HbA1c measurement, lipid profile, and body composition, as well as register of insulin dose, pre- and post-meal capillary glucose, and frequency of hypoglycemia. After this, Lina- gliptin was started at dose of 5mg every 24 h during 3 months. Patients had a monthly follow-up visit and the adjustment of insulin dose was performed by a blinded Endocrinologist. At 3 months, all the basal measurements were repeated. Data were analyzed with a paired t test (p less than 0.05). Results: Seven patients were included in the study but only 5 (4 female and 1 male) completed the 3 months follow-up. The patient´s average age Therapeutics and duration of the disease were 21 ± 6 and 9.6 ± 5.5 years, respectively. Clinical Diabetes/ PUBLISHED ONLY There was a decrease in HbA1c (10.2 ± 1.9 vs. 9.3 ± 0.3%, p 0.257), pre-meal Supported By: Pennington Biomedical Research Foundation; National Institute glucose (169.3 ± 16.7 vs. 154.6 ± 0.4 mg/dl, p 0.427), post-meal glucose (174.8 of General Medical Sciences (1U54GM104940); Nutrition Obesity Research Cen- ± 13.3 vs. 155.3 ± 13.9 mg/dl, p 0.014), insulin dose (40.6 ± 9 vs. 36.6 ± 13.3 ters (P30DK072476); Eunice Kennedy Shriver National Institute of Child Health and IU, p 0.036), and frequency of pre and post-meal hypoglycemia (3 vs. 2, and 2 Human Development (1UG1HD090967 to D.S.H.) vs. less than 1 per month, respectively, p 0.215). Conclusion: Linagliptin added to insulin scheme improved metabolic con- 2457‑PUB trol and reduced the frequency of hypoglycemia; DPP-IV inhibitors and other The Efficacy of Pediatric Outreach Diabetes Clinics vs. Central Dia‑ oral therapies should be more evaluated to define the role they may have as betes Clinics: A Retrospective Chart Review Study a complementary treatment to insulin in uncontrolled patients with type 1 ROB LINDSAY, DANIA AL-HAMAD, Salt Lake City, UT diabetes. Little has been reported about the efficacy of pediatric outreach (OR) Supported By: Hospital Regional de Alta Especialidad del Bajio, León, Guana- diabetes clinics in general, and more specifically in pediatrics. Primary Chil- juato, México dren’s Hospital provides care to children from 6 states in addition to Utah and has had an OR program since 1982. This study compares one provider’s 2459‑PUB patients’ Hemoglobin A1c levels from OR and the central clinic over the past Fifty Shades of Transferring Pediatric Diabetes Patients to Adult 8 years to assess the quality of care provided to these distant patients. Inter- Care estingly, there was no significant difference between any of the groups, con- SARA MOASSESFAR, SALEH ADI, STEPHEN E. GITELMAN, MEGUMI OKUMURA, sidering that the central patients had access to clinic social workers, more MARCELA ARREGUI REYES, San Francisco, CA flexibility in scheduling and rescheduling and more opportunity to meet with Background: Studies show youth “graduating” out of pediatric diabetes diabetes educators and dietitians between visits. This study shows that dia- (DM) care have delayed follow-up in adult clinics. To counter this, our pedi- betes control tends to be similar in OR and central settings. atric DM center developed a transition program to prepare patients for self- Table. Male Diabetes Patients Hemoglobin A1C Central Clinics vs. OR management and help establish adult care. No study has explored reasons (n=Number of Encounter). why young adults transfer out of pediatric care, thus we sought to explore Central OR these and the prevalence of establishing adult care. Methods: We compared 2 groups of adults transferring out of our care in Age n A1C% n A1C% the past 3 yrs. Group 1 preceded our transition program (n=21) and Group 1 visit within 1 yr of transfer, b) if yes, when 12-15Y 1147 8.8 994 8.6 and where, c) if no, why not and d) reasons for transfer. Nonresponders and those responding but not in adult care were recontacted in 3 mos and left up 16-18Y 680 8.5 667 8.6 to 2 messages. Descriptive statistics and t-tests were performed on base- Total 3073 8.4 2953 8.3 line characteristics and outcomes. Results: Baseline characteristics were similar between groups. In the Table. Female Diabetes Patients Hemoglobin A1C Central Clinics vs. OR combined cohort, age was 18-28 yrs (mean 20.6); 49% was female; 81% (n=Number of Encounter). Caucasian; 33% publicly insured; 90% had T1D; 10% T2D, CFRD or MODY; Central OR mean DM duration was 11 yrs; and mean HbA1c was 9.0%. For Group 1, Age n A1C% n A1C% 7/21 patients (33%) responded, of which 6 (85.7%) had established adult care within 1 yr of transfer. For Group 2, 18/36 responded, of which 13 (72%)

ADA-Supported Research For author disclosure information, see page A751.

A641 PEDIATRICS—TYPECATEGORY 1 DIABETES

involved in our transition program had higher rates of responding, sharing 2462‑PUB clinical information, and establishing adult care or being in the process. Construction of a Diabetes Well-Being Survey for Teens with Type 1 Supported By: National Institutes of Health (K12DK094726) Diabetes (T1D) BARBARA J. ANDERSON, CHARLOTTE COHEN, WENDY LEVY, DEMET OZKAYA, 2460‑PUB THOMAS DANNE, LORI M. LAFFEL, Houston, TX, Boston, MA, Paris, France, Hannover, Diabetes Management Difficulties and Strategies in Children with Germany Autism Spectrum Disorder and Type 1 Diabetes Objectives: Although teens with T1D are vulnerable to depression, dia- KELLY STANEK, ERIN YOUNGKIN, JENNIFER K. RAYMOND, SHIDEH MAJIDI, betes distress and reduced quality of life, no single brief measure assesses Aurora, CO, Los Angeles, CA their diabetes well-being. Patients with Autism Spectrum Disorder and type 1 diabetes (ASD+T1D) Methods: An international multidisciplinary panel (peds, endos, psycholo- may encounter unique difficulties managing T1D and distinct strategies may gists, RNs/RDs) informed design of a well-being survey for teens with T1D. improve T1D care. A questionnaire on independent tasks, difficulties, and The survey, including a 5-point Likert response scale, had 6 domains, each strategies in T1D management was given to caregivers of patients with with 2 positive and 2 negative items: self-worth; diabetes distress/worry; ASD+T1D (n=13) and a control group of T1D patients without ASD (n=39) communication; social/peer integration; weight/body image; and family. To matched for age, sex (92% male), race/ethnicity, and T1D duration. More assess clarity and acceptability of the 24-item survey and to identify 12 pre- ASD+T1D caregivers indicated their child could not complete any T1D care ferred items (2 for each domain), 18 younger (13-15 yrs) and 18 older teens task independently (30.8% ASD+T1D vs. 0% T1D). Both groups ranked spe- (16-18 yrs) with wide ranging A1c values (<7->10%) completed the survey cific difficulties from most to least difficult. Change in diet (46% vs. 44%) (completion time 5-10 min) and then responded to a standardized cognitive followed by carb counting (24% vs. 31%) were the 2 most difficult tasks in interview by trained research staff at 2 U.S. diabetes centers. Interviews ASD+T1D and T1D cohorts. Less than half of the ASD+T1D group (46.2%) were audio-recorded and analyzed for common themes about clarity, accept- had a strategy to overcome difficulties, compared to 87.2% of controls. Both ability and item preference. groups selected insulin pump as the most useful strategy (23.1% ASD+T1D Results: Teens (50% male) had mean age 16, T1D duration 10 yrs; most vs. 46.2% T1D). ASD+T1D families preferred using schedules (23.1%) while were pump-treated. There was strong consensus that items about weight T1D families used reminders (43.6%). Overall, controls appeared to have and peer relations were not relevant or acceptable, whereas items about greater ability to perform diabetes care tasks independently and have family interactions, motivation and support for T1D self-care were most strategies to overcome difficulties in management. Although some with important. Teens preferred positively to negatively worded items and Therapeutics ASD +T1D may be unable to gain complete independence, development and reported feeling comfortable sharing responses with their healthcare teams Clinical Diabetes/ to benefit clinical interactions and inform recommendations. In response to PUBLISHED ONLY discussion of strategies to overcome difficulties may be helpful for T1D management in patients with ASD+T1D. a strong preference for items about family, motivation and support, we created a revised 12-item well-being survey reflecting teen recommendations. Table. Conclusions: While teens struggle to attain A1c targets, assessing teen ASD+T1D Control well-being may offer opportunities to intervene on modifiable family factors Age 12.62 ± 0.88, 12.41 ± 0.46, related to provision of family support and motivation for self-care. A valid n=13 n=39 and reliable survey of teen diabetes well-being emphasizing family issues Independent Tasks may yield an important patient-reported outcome. Check BG independently 9 (69.2%) 39 (100.0%) Supported By: Sanofi Count Carb independently 4 (30.8%) 29 (74.4%) Insulin Dosing 5 (38.5%) 22 (56.4%) Give Insulin independently 6 (46.2%) 17 (43.6%) 2463‑PUB Easing the Transition: A Program for College-Bound Teens with Presence of Strategy 6 (46.2%) 34 (87.2%) Diabetes Number of Strategies to overcome difficulties 0.85 ± 0.27 1.44 ± 0.13 JULIE M. SURHIGH, DAVID P. TOBIN, Royal Oak, MI, Rochester, MI Background: This study aimed to evaluate the effectiveness of a pre- 2461‑PUB college education program at improving knowledge, self-care skills and glycemic control in graduating high school seniors with type 1 diabetes. A Parental Views on Transitioning to Adult Management in Patients one-time education program was piloted by our division in May 2015. with Autism Spectrum Disorder and Type 1 Diabetes Subjects and Methods: 60 high school seniors with diabetes were invited SHIDEH MAJIDI, KELLY STANEK, ERIN YOUNGKIN, JENNIFER K. RAYMOND, to attend. The program consisted of a lecture followed by a discussion fea- Aurora, CO, Los Angeles, CA turing a panel of college students with type 1 diabetes. Educational curricu- Transferring to adult diabetes care is a challenge for patients with type 1 lum was adapted from the American Association of Diabetes Educators’ 7 diabetes (T1D) and may be more difficult for those with Autism Spectrum Self-Care Behaviors. Participants completed a pre-class quiz, an immediate Disorder (ASD) and T1D. Resources for patients with Autism Spectrum Dis- post-class quiz, and a 1-year follow-up quiz. Of the 60 students invited, 12 order and type 1 diabetes (ASD+T1D) transitioning from pediatric to adult attended the program and 7 completed the follow-up quiz at 1 year and were providers are lacking. Via a questionnaire completed by caregivers, our study enrolled in the study. examined views and comfort with transitioning in those with ASD+T1D 12 ≥ Results: Of the 7 high school seniors enrolled in the study, 57.1% used years of age (n=6) compared to a control group of T1D patients (n=28). The insulin injection therapy and 42.9% used insulin pump therapy. Mean HbA1c average age of ASD+T1D patients was 15.5 years (range 13-17 years) vs. was 8.04±0.92%. There was a significant improvement in mean quiz score 15.2 years (range 12-18; one patient 12 years). Both groups believed transi- after attending the program (P=0.0113), with an average improvement in tion should occur at an average of 18.5 years. More families with ASD+T1D score of 2.14±1.57 points (maximum score 20 points). Significant improve- had started thinking or creating a transition plan (66.7% ASD+T1D vs. 42.9% ment in mean quiz score was maintained at one year (P=0.0353), with an T1D), and few providers had discussed transitioning with families (16.7% average improvement in score of 1.43±1.40 points. Quiz questions most ASD+T1D; 14.3% T1D). Both ASD+T1D and T1D families wanted a provider frequently missed prior to the program included topics on alcohol consump- to discuss a plan with them (100%; 70.9%). Families with ASD+T1D feel less tion, mental health, and understanding how a correction factor works. There confident about receiving adequate information and tools for transitioning was no significant change in HbA1c values between baseline and one-year (66.7% ASD+T1D very or somewhat confident vs. 85.8% T1D). ASD+T1D follow-up. families were most concerned about switching care facilities and less per- Conclusions: Our data suggest that knowledge gaps exist in specific areas sonal interaction when transitioning to adult care, while T1D families were of diabetes self-care unique to young adults transitioning to college. A one- most concerned about their child’s medical response to change in care (i.e., time education program offered to graduating high school seniors appears changes in A1c level and self-management). Overall, we found additional to improve short- and long-term knowledge of diabetes self-care. Glycemic guidance regarding transition for all patients with T1D is needed. Those with control was not significantly altered; however, further investigation utilizing a comorbid condition may start formulating a transition plan at a younger a control group would help to better elucidate the impact of the program. age, indicating that diabetes providers should initiate transition conversa- tions earlier for those with additional diagnoses, such as ASD. Larger studies may help determine the specific resources needed.

For author disclosure information, see page A751. ADA-Supported Research

A642 PEDIATRICS—TYPECATEGORY 1 DIABETES

2464‑PUB 2466‑PUB Professional Continuous Glucose Monitoring before Starting Insu‑ Changing Profile of GAD and IA-2 Antibody Positivity in Recently lin Pump Therapy Diagnosed Type 1 Diabetes Indian Children JESSICA A. FERRIS, ROSHANAK MONZAVI, Los Angeles, CA SUDIP CHATTERJEE, DEBMALYA SANYAL, SANDIP K. BATABYAL, SUBRATA When patients with type 1 diabetes (T1D) transition from a multiple daily MAITY, Kolkata, India injection regimen to insulin pump, glycemic control may initially worsen until Objective: Existing evidence on type 1 diabetes (T1D) patients from India optimal doses are found. Continuous glucose monitoring (CGM) provides proposes that fairly large number are negative to GAD 65 (Glutamic Acid detailed blood glucose (BG) data, which can be used to personalize starting Decarboxylase) and IA-2 (Insulinoma Antigen 2) auto antibodies. Socio-eco- pump insulin doses. We conducted a randomized controlled pilot study using nomic factors and enterobiome changes are well known to alter antibody Medtronic iPro2 CGM for insulin dose determination at pump initiation in positivity rates. We wanted to see whether there were secular changes in T1D youth to assess whether its use would improve time within target BG antibody positivity rates among newly diagnosed T1D children in response (70-180 mg/dL). Ten subjects aged 4-16 years with T1D for ≥3 months, naïve to rapid transition in Indian lifestyle. to CGM, and starting on insulin pump were recruited. All subjects wore the Methods: All newly diagnosed T1D patients aged less than 16 years had CGM for 5 days at ~2 weeks before, initial week, and 6 weeks after pump GAD and IA-2 antibodies measured between 2007 and 31st August 2016. start. In the intervention group (INT), CGM data was used to determine insu- Results: The chi-square test for trend analysis showed a significant rising lin pump doses. Subjects were 10.6±3.8 years old and had T1D for 2.5±3.5 trend for IA2 antibody alone positive (p<0.001, chi-square for trend=17.437, years. No differences were noted between groups at baseline: % time in df=1) and either antibody positive percentages (p<0.001, chi-square for target BG (63.2 vs. 63.5%), mean BG (173 vs. 167 mg/dL), and MAGE (108 vs. trend=22.71, df=1), but not in the GAD antibody positivity (p=0.059, chi- 90.5 mg/dL) for INT vs. control group, respectively. Preliminary data showed square for trend=3.567, df=1) and in dual antibody positive percentages a trend towards improved glycemic control in INT and worsened control in (p=0.486, chi-square for trend=0.485, df=1) over a period of ten years i.e., the control group. Differences between groups were not statistically sig- from 2007 to 2016. nificant, most likely due to insufficient power. Using professional CGM prior Conclusion: There is a substantial change in the antibody positivity rates to insulin pump appears to improve glycemic control and decrease glycemic in newly diagnosed T1D children over the last nine years. We postulate the variability during transition to pump therapy. increase in autoimmunity is attributed to the changes in the enterobiome, Table. obesity (BMI), hygiene, socioeconomic development and exposure to oral polio vaccine. This rise in autoimmunity may also be a significant contribut- Intervention Control ing factor towards the recent increased incidence of T1D in India. Therapeutics

Group Group Clinical Diabetes/ PUBLISHED ONLY % time in target - 5 days 71±8.2 57.5±10 2467‑PUB % time in target - 6 weeks 81.7±13 53.5±13 Disordered Eating Develops Early in Adolescents with Type 1 Dia‑ Mean BG (mg/dL) - 5 days 146±20 167±25 betes Mean BG (mg/dL) - 6 weeks 137±6.6 203±71 LISAL J. FOLSOM, TAMARA S. HANNON, Indianapolis, IN Mean Amplitude of Glycemic Excursion (mg/dL) - 5 days 93.1±26 102±30 Background: There is an increased incidence of both disordered eating behaviors (DEB) and eating disorders (ED) in adolescents with T1D. Patients Mean Amplitude of Glycemic Excursion (mg/dL) - 6 weeks 61.5±7.0 127±14 with T1D and concurrent DEB or ED have more medical complications and a tripled mortality rate. There would be great benefit in determining risk fac- Supported By: Medtronic; Children’s Hospital Los Angeles tors for DEB and ED, and intervening before these harmful behaviors arise. Objectives: To establish the course of eating behaviors in adolescents 2465‑PUB with newly diagnosed T1D (NDT1D) and determine if change in these behav- Percentage of Type 1 Diabetes for Children and Adolescents in Tai‑ iors negatively affects glycemic control. wan: A Nationwide Data Methods: Subjects with NDT1D were followed prospectively for 6 months YI DER JIANG, SHIN YU LIEN, Taipei, Taiwan, Taoyuan, Taiwan after diagnosis. BMI and HbA1c were obtained at each visit. Validated ques- Background: The urine screening program for nearly 3 million schoolchil- tionnaires and online dietary recall software were used to measure eating dren was the first nationwide data that type 2 diabetes was predominant behaviors and nutrition knowledge. Descriptive statistics, student’s t-tests, for schoolchildren but limited by 38% with unknown type. We tried National and regression analyses were used for analysis. Health Insurance (NHI) data to demonstrate the percentage of type 1 diabe- Results: Twenty-five patients with NDT1D participated. Age ranged tes aged 5 to 19 y/o instead. from 10-18 years. HbA1c was 13.5 ± 2.2% at diagnosis, and 6.8 ± 1.2% Methods: The case number of ICD-9-CM code 250, the combination of at 6 months (p<0.01). Average BMI increased by 2.6 ± 0.8 kg/m2 (p<0.01). type 1 and type 2 diabetes, in either inpatient or outpatient diagnosis, was Reported intake of total calories, carbohydrate, sugar, fat and protein did the official data announced annually by Ministry of Health and Welfare in not change, however nutrition knowledge significantly increased (p=0.05). Taiwan, together with that of NHI insured population. Type 1 diabetes was Female subjects were more likely to screen positively for DEB than males at based on the issue of the Catastrophic Illness Certificate from NHI with the 6 months (p=0.02). Positive screening for DEB correlated with a higher HbA1c same code. In addition of NHI coverage rate, the percentage of type 1 dia- at 6 months (p=0.03, R2=0.24). betes among schoolchildren 5~19 years of age can be shown from 2003 to Conclusion: Although nutrition knowledge increased, dietary composi- 2010, and stratified by gender and 5-year age groups. tion did not change. Positive screening for DEB was associated with higher Results: The NHI coverage rate varied between 97% and 98.4%, high HbA1c. This finding supports the hypothesis that patients with DEB have enough for a nationwide data. The percentage of type 1 diabetes ranged poorer glycemic control, and suggests this trend manifests as early as 6 from 26% to 29.9% with mean value of 28.1% from 2003 to 2010. It was months after diagnosis. Our results indicate that targeted nutrition educa- higher for girls (30.9%) than that for boys (25.6%) in each age group, and tion is critical in T1D, as reported dietary intake did not change despite nutri- negatively associated with age for both gender. tion education. They also highlight the need for interventions that translate Conclusion: In urine screening program, type 1 diabetes (n=78) accounted nutrition knowledge into practical methods that adolescents will embrace. for 13.4% of all schoolchildren diabetes (n=581), and 21.6% after exclud- Supported By: National Institutes of Health (2T32DK065549) ing 230 unknown type or without physician’s diagnosis of diabetes. If only type 1 and type 2 (n=257) diabetes included, type 1 diabetes accounted for 2468‑PUB 23.3%, very close to 26.0 ~ 29.9% in this study, but much lower than over Does Iron Status Explain Racial Disparity in Mean Blood Glucose 90% in western countries. Such a marked difference may come from low (MBG) Independent of HbA1c Outcome for Children with Type 1 DM incidence of type 1 diabetes and high prevalence of impaired fasting glucose (T1D)? (IFG). Incidence of type 1 diabetes in Asia was no more than 5, much lower MAHMOUD A.A. HAMDAN, STUART CHALEW, New Orleans, LA than 10~65 per 100,000 in western countries. IFG for adolescents was 22% Introduction: Blacks (B) have higher HbA1c than whites (W) independent of in Taiwan, and correlated to male, obesity, high blood pressure, high triglyc- differences in MBG and RDW-CV. Decreased iron status is associated with eride or low HDL. An intervention program must be designed for boys to increased HbA1c. Thus low iron may account for higher HbA1c. prevent type 2 diabetes. Methods: We evaluated iron status in a mixed race sample of youth with Supported By: Taiwan Ministry of Science and Technology T1D. Labs were obtained for ferritin, soluble transferrin receptor (sTfR),

ADA-Supported Research For author disclosure information, see page A751.

A643 PREGNANCY—CLINICAL/EPIDEMIOLOGYCATEGORY

HbA1c and CBC. MBG was derived from glucose meter records over last 30 new long acting insulin associated with low blood glucose variability. Real days. Statistical analysis was done by t-test and multiple variable analysis. life data in adults with T1DM and T2DM demonstrated significant reductions Results: Comparison of variables between groups by t-test presented in in HbA1c, hypoglycemic episodes, and insulin dose. Our aim was to evalu- Table. HbA1c and MBG were higher in B than W. Ferritin was correlated ate glycemic control following switching to Tregludec in children and young with Hb but not HbA1c or MBG. Differences in ferritin and sTfR between adults. We included patients <31 years diagnosed with T1DM with HbA1c the groups were not significant. WBC and platelet levels were not differ- >7.5 who were switched to Tregludec. HbA1c was measured at baseline and ent between groups. In multiple variable analysis race (p=0.0241) and MBG 3-6 months after the switch. The primary end point was change in HbA1c. (p=0.0042) were statistically influential on HbA1c, r2=0.34, p=0.0002. Fe, Dose of basal and total insulin (IU/kg body weight/day) were calculated. Self- sTfR or ratio were not statistically significant when added to the model. monitoring of blood glucose measurements were documented two weeks Conclusion: After adjustment for race and MBG, iron indices were not before and 3 months after the switch. 90 patients (56% girls, age range independent predictors of HbA1c levels. These observations indicate that 2.9-31 years, mean 14±5.4 years) were included. Mean duration of T1DM factors besides iron status contribute to MBG-independent racial disparity was 5.5±4.8 years. Previous basal insulin treatments were Lantus (68%), in HbA1c between B and W pediatric patients. Levemir (25%), Toujeo (1.9%), or CSII (5.1%). Mean HbA1c at baseline was 9.04±1.7. After 3 months of Tregludec treatment mean HbA1c decreased to 8.47±1.6 (p<0.0001). After 6 months it decreased further to 8.23±1.3 (p<0.02). Mean basal and total insulin dose at baseline were 0.38±0.17 IU/kg/day and 0.87±0.32 IU/kg/day, respectively. After 3 months of Tregludec treatment mean basal and total insulin dose decreased to 0.35±0.13 IU/kg/day and 0.76±0.23 IU/kg/day (p<0.0001, p<0. 01), respectively. After 3 months mean glucose measurements decreased from 204.6±68.5 to 189.7±61 mg/ dl (p<0.03) and the percent of patients with very high (>300 mg/dl) and very low (<50 mg/dl) glucose levels decreased significantly (p<0.015, and p<0.017, respectively). Less hypoglycemic episodes were observed after the switch. These results demonstrate that switching to Tregludec as basal insulin in children and young adults with T1DM improves glycemic control. Supported By: Endocrine Fellows Foundation Therapeutics

Clinical Diabetes/ PREGNANCY—CLINICAL/EPIDEMIOLOGY PUBLISHED ONLY 2469‑PUB Oral Administration of Lactobacillus Reuteri Improves Periodontal Disease in Children and Adolescents with Type 1 Diabetes 2471‑PUB BARBARA PREDIERI, SARA BARBIERI, DALILA MICELI, CHIARA CATTELANI, SILVIA MAZZONI, ANDREA FORABOSCO, LORENZO IUGHETTI, Modena, Italy WITHDRAWN Objectives: There is disagreement on the effect of diabetes on oral hygiene. Probiotics creating a biofilm could protect the oral tissues against periodontal bacteria. We aimed to assess the effects of Lactobacillus reuteri administration upon the oral health of children and adolescents with type 1 diabetes (T1D). Methods: Forty-three patients (11.3±2.77 yrs.; T1D duration 58.2±38.0 months) were enrolled and randomly assigned to Group A (probiotic - 10^8 CFU/day for 3 months) and Group B (no probiotic). Full Mouth Plaque Score (FMPS), Full Mouth Bleeding Score (FMBS), insulin dose (IU/kg/day), and HbA1c were measured at baseline (T0) and 3-months after (T1). Results: FMPS significantly improved in both Group A and B (p<0.05). In Group B insulin dose increased (p=0.01) and HbA1c improved (p<0.001) at T1; in Group A the metabolic control was unchanged. In Group A, 13 out 22 patients reported a regular probiotic intake (Group A1), while the other ones used it sporadically (Group A2). Despite FMPS and FMBS values were not different between groups at T0, they were significantly lower (p<0.05) in Group A1 respect to Group A2 at T1 and longitudinally decreased in Group A1 but not in Group A2 (Table). Conclusions: Our preliminary data suggest that 3-months administration of probiotic might improve the oral health of children and adolescents with T1D and confirm the influence of the glycemic control. Table. Data are Reported as Mean ± SD (Median). Group A1 Group A2 p T0 FMPS (%) 88.8 ± 11.2 (90.0) 92.8 ± 12.5 (100.0) 0.367 FMBS (%) 64.6 ± 28.7 (60.0) 78.3 ± 24.1 (80.0) 0.317 T1 FMPS (%) 70.0 ± 27.1 (80.0) 86.1 ± 27.5 (100.0) 0.038 2472‑PUB p=0.017 vs. T0 p=0.465 vs. T0 Bone Turnover in Women with Prior Gestational Diabetes Mellitus Before and After 52 Weeks’ Treatment with Liraglutide FMBS (%) 49.6 ± 23.7 (50.0) 78.9 ± 28.4 (95.0) 0.025 LOUISE VEDTOFTE, ELISABETH HANNERUP, SIGNE FOGHSGAARD, THORA BUHL, p=0.074 vs. T0 p=0.787 vs. T0 ELISABETH R. MATHIESEN, JENS A. SVARE, TINE D. CLAUSEN, PETER DAMM, NIKLAS R. JØRGENSEN, FILIP K. KNOP, TINA VILSBØLL, Hellerup, Denmark, 2470‑PUB Copenhagen, Denmark, Herlev, Denmark, Hillerød, Denmark Real-Life Data: Insulin Tregludec in Children and Young Adults with Liraglutide induces weight loss and as weight loss is associated with loss T1DM of bone mass, we investigated the effect of liraglutide-treatment in women NEHAMA ZUCKERMAN-LEVIN, SHALEV ZUCKERMAN, NAIM SHEHADEH, Haifa, with prior gestational diabetes mellitus (pGDM) on whole-body bone mineral Israel density (BMD), whole-body bone mineral content (BMC), the bone resorption Achieving good glycemic control is a major goal in T1DM. This includes marker C-terminal telopeptide of type 1 collagen (CTX-I) and the bone for- striving for a target HbA1c, weight adjusted insulin dosing, minimizing hypo- mation marker procollagen type 1 amino-terminal propeptide (P1NP). One- glycemic episodes, and decreasing blood glucose variability. Tregludec is a hundred and two women with pGDM (age: 38±5 years (mean±SD), BMI: 32±4

For author disclosure information, see page A751. ADA-Supported Research

A644 PREGNANCY—CLINICAL/EPIDEMIOLOGYCATEGORY kg/m2) underwent a 4-hour 75-g oral glucose tolerance test at baseline and post-prandial measurements 100-140 minutes after meals. An inadequate after one year’s randomised and blinded intervention with placebo (n=54) or delay was associated with origin and a higher HbA1c at inclusion (5.3±0.5 vs. liraglutide, 1.8 mg once-daily, (n=48). During the trial, 10 women withdrew 5.1±0.4%, p=0.03). A poor concordance (23.1% of women had <90% matched from the liraglutide group and 9 from the placebo group. Fifteen controls values in diary and meter memory) was associated with family history of matched for age and BMI were studied once (age: 39±4 years, BMI: 31±5 kg/ diabetes; 9.9% of women were considered as dissemblers as they underesti- m2). Dual energy X-ray absorptiometry was used to assess whole-body BMD mated glucose values or did not report a high glucose value more than three and whole-body BMC. At baseline, there were no differences between con- times a week. Poor adherence was associated with more preeclampsia (12.2 trols and women with pGDM with respect to whole-body BMD, whole-body vs. 1.9%, p=0.049) and inadequate post-prandial delay with a high HbA1c BMC and the total area under the curve (tAUC) during the OGTT for CTX-I at delivery (5.3±0.4 vs. 5.0±0.3%, p<0.01), despite a higher rate of insulin and P1NP. Compared to placebo, liraglutide-treatment induced a significantly therapy. To conclude, although women with GDM are considered as highly greater weight loss after 52 weeks (4.9±6 vs. 1.3±5 kg, P<0.005). No changes motivated, poor SMBG adherence and reliability raise concern and are asso- in whole-body BMD, whole-body BMC, tAUC for CTX-I or tAUC for P1NP ciated with poor prognosis. were observed in the two groups. We conclude that liraglutide-treatment- Supported By: Roche Diagnostic France induced weight loss in women with pGDM seems not to affect bone turnover, whole-body BMD and whole-body BMC respectively, suggesting that 2475‑PUB liraglutide treatment from a bone turnover perspective is safe to use in this Does Early Gestational Diabetes Mellitus Correspond to Unknown population. Prediabetes Before Pregnancy? Arguments from Metabolic Mark‑ Supported By: Novo Nordisk A/S ers at Initial Care EMMANUEL COSSON, LUCIO BIANCHI, FRANÇOISE GARY, DORIAN SANDRE- 2473‑PUB BANON, YAHYA JABER, ISABELA BANU, CAMILLE PILLEGAND, SABRINA CHI- Families Defeating Diabetes (FDD): Focus Group Analysis of a Post‑ HEB, MINH TUAN NGUYEN, PAUL VALENSI, Bondy, France partum Healthy Living Intervention for Women with Gestational Screening for dysglycemia at first antenatal visit was proposed to detect Diabetes (GDM) unknown diabetes apart from pregnancy (Diabetes in pregnancy: DIP). Early TANIYA NAGPAL, MICHELLE MOTTOLA, RUTH M. MCMANUS, London, ON, gestational diabetes mellitus (eGDM), beginning before pregnancy-induced Canada insulin resistance, might correspond to unknown prediabetes. From a series FDD was a postpartum Canadian diabetes healthy living intervention for of 188 pregnant women without known diabetes before pregnancy, we con- women with recent GDM within their family context. We report results from sidered 126 women having had skin forearm skin autofluorescence measure Therapeutics Clinical Diabetes/

a Focus Group program analysis. (AGE Reader (tm)): 8 women without eGDM and regular GDM (IADPSCG PUBLISHED ONLY Women (W) with recent GDM and interested partners (P) participated in criteria); 62 women with eGDM (<22 weeks of gestation); 48 women with FDD, provided from months 3 to 12. Study components included: DM pre- regular GDM but normal early screening; and 8 women with DIP. Women vention seminar; password-protected website; automatic email hints 2X/ with dysglycemia also had fasting plasma glucose, insulin, fructosamine month; weekly walking group. W and P finishing 12 month follow-up were and HbA1c measurements when they were referred for care. Skin autofluo- invited to a Focus Group to assess perceived value of the program compo- rescence (advanced glycation endproducts (AGE), arbitary units) differed nents. Responses were recorded anonymously; thematic analysis was done according to glycemic status (p<0.05 after adjustment for age): no GDM identifying most prominent comments; results are reported qualitatively. (1.79±0.32), regular GDM (1.99±0.47), eGDM (2.11±0.48) and DIP (2.42±0.34); Seven people attended (4W, 3P) a 3 hour session. Comments: Walk- with different proportions of women having AGE >1 standard deviation (SD) ing group: opportunity to socialize with other W (4/4W); should last past (41.7; 45.8; 54.8 and 100%, respectively; p<0.05) or >2 SD for age (8.3; 10.4; 12 months (4/4W); allowed partner to get out of house (3/3 P). Nutrition: 25.8 and 50%, respectively; p<0.05). Considering only women with regular assisted healthy choices (4/4W); P felt most engaged in diet component, GDM, eGDM and DIP, fructosamine (196±12; 203±18 and 223±35 µmol/L, aided them to make healthier choices (2/3P). P perceived their biggest study respectively, p<0.0001), and HbA1c levels (4.9±0.5; 5.0±0.4 and 5.8±0.8%; role was making healthy food choices (3/3P); 2/3P had not realized their diet p<0.001) differed significantly but not HOMA-IR index (2.73±1.59; 2.78±1.76 was unhealthy until FDD; 2/4W reported being pleased P was learning about and 4.39±2.41; p=0.057). To conclude, fructosamine and HbA1c are high in diet as this supported their choices. Message vectors: seminar was informa- case of DIP but not of eGDM. Skin autofluorescence is higher in women with tive for diet/breastfeeding information (4/4W); 3 months postpartum was eGDM and DIP than in those without GDM and regular GDM, suggesting appropriate timing (4/4W); email hints 2X/month best for messaging (4/4W). that these conditions actually correspond to dysglycemia before pregnancy. W and P preferred receiving short email hints, none found website useful. All P said seminar helpful but preferred email provision vs. face-to-face. P 2476‑PUB had no interest in any social media options. All W preferred receiving study Assessment of Insulin Resistance and Secretion Indexes in the hints at least twice a week. Nonpregnant Condition as Predictors for Insulin Use of Gestational Focus Group analysis of the FDD healthy living program for women with Diabetes Mellitus recent GDM and interested partners concluded access to a study website KYOKO KOHASHI, YUSAKU MORI, ANNA OSAMURA, MASAKO TOMOYASU, was not of value; whereas helpful study components included: postpartum YUKI TANABE, TOMOYASU FUKUI, TSUTOMU HIRANO, Tokyo, Japan seminar at 3 months; weekly walking group; frequent automatic provision of Background: Gestational diabetes mellitus (GDM) patients at high risk for healthy living hints by email. insulin therapy need careful follow-up. We examined whether insulin secre- Supported By: International Diabetes Federation tion and resistance in the non-pregnant condition are useful predictors of insulin use in GDM. 2474‑PUB Methods: GDM patients (all Japanese) admitted to Showa University Hos- Poor Adherence to and Reliability of Self-Monitored Blood Glucose pital between December 2004 and February 2016 were recruited retrospec- in Women with Gestational Diabetes Mellitus Are Usual and Asso‑ tively. Insulin therapy was initiated when a self-monitored blood glucose ciated with Poor Outcomes level over 100 mg/dL in the fasting state or 120 mg/dL 2 hours after a meal EMMANUEL COSSON, BAZ BAZ, FRANÇOISE GARY, DORIAN SANDRE-BANON, was consecutively observed. Insulin secretion indexes (ISIs: insulinogenic ISABELA BANU, SABRINA CHIHEB, YAHYA JABER, PAUL VALENSI, Bondy, France index [II], homeostasis model assessment [HOMA]-β, and area under the We aimed to evaluate adherence to self-monitored blood glucose (SMBG) curve of insulin/glucose [AUC I/G]), insulin resistance indexes (IRIs: HOMA-R and reliability of blood glucose diary records in women with gestational dia- and composite index [CI]), and disposition index (DI) were calculated from a betes mellitus (GDM), their determinants and prognosis. We prospectively 75-g oral glucose tolerance test conducted 6 to 8 weeks after delivery. The selected women with newly-diagnosed GDM who were referred to our dia- odds ratio per unit change is shown with a 95% confidence interval. betes management program, spoke national language and had understood Results: Insulin-treated GDM (111 patients out of 393) showed higher glycaemic SMBG techniques and goals. During the first follow-up visit, we HOMA-R and lower CI in contrast to similar ISIs. In univariable logistic collected SMBG results from glucometers and diary records. We analyzed regression, IRIs, but not ISIs, were associated with insulin use. In multiple pregnancy outcomes. Data were analyzed over 13±3 days in 91 women. Only logistic regression with the simultaneous method, DI or a combination of 61.5% of them had performed >80% of required measurements. Poor adher- one IRI and one ISI was used in addition to diabetes family history, age, ence was associated with family history of diabetes, social deprivation and gestational week at diagnosis, body mass index prior to pregnancy, and non-European origin. Average delay between pre- and post-prandial mea- body weight change during the gestational period. Although the model using surements was 131±26 minutes; with 46.5% of women performing 80% of HOMA-R and -β showed the highest coefficient of determination (R^2), the

ADA-Supported Research For author disclosure information, see page A751.

A645 PREGNANCY—CLINICAL/EPIDEMIOLOGYCATEGORY

value was only 0.092 (p<0.001; HOMA-R: 1.95 [1.20-3.16], p<0.01; HOMA-β: and important in predicting the risk of future prediabetes. The small numbers 0.99 [0.98-1.00], p=0.02). however suggest the need for larger scale studies to confirm our findings. Conclusion: Although insulin resistance and secretion assessed as HOMA-R and -β in the non-pregnant condition are predictors for insulin use 2479‑PUB in GDM, their clinical usefulness is limited. The Prevalence of Gestational Diabetes in the Province of Trento, Trend 2010-2015 and Adverse Neonatal Outcomes 2477‑PUB SILVANO PIFFER, RICCARDO PERTILE, TIZIANA ROMANELLI, Trento, Italy Gestational and Pregestational Diabetes Exposure In Utero: Valida‑ Introduction: A blood glucose control at the initial assessment (before the tion of a Definition for Use in Administrative Data 13th week of gestation) and at a second check between the 24-27 week is ELIZABETH A.C. SELLERS, RANDALL FRANSOO, DAN CHATEAU, HEATHER recommended in Italy for the monitoring of physiological pregnancy. These PRIOR, ANGELA TAN, HUI CHEN, BRANDY A. WICKLOW, Winnipeg, MB, Canada data as all data concerning the characteristics of the mother, the care during Background: There are reported differences in the impact of gestational pregnancy and childbirth are registered in the medical birth register of the (GDM) vs. pregestational diabetes (PGDM) exposure on the metabolic health province of Trento. of offspring making differentiation imperative. Aims: To estimate the prevalence of gestational diabetes mellitus (GDM) Aim: To determine the best administrative data case definitions for GDM in pregnant women assisted at the birth centers of Trento Province from and PGDM in pregnancy. 2010 to 2015, verifying the differences in relation to citizenship and the Methods: We compared the performance of several case definitions for impact on the newborns. GDM and PGDM within the administrative health data housed in the Mani- Methods: A retrospective birth cohorts study was conducted for the years toba Population Health Research Repository at the Manitoba Centre for 2010-2015, including a total of 29.025 pregnant women. The annual GDM Health Policy to an identified population of women in whom the diagnosis prevalence was calculated analyzing the data according to nationality and of GDM or PGDM was known from the clinical database of the Diabetes whether the foreign, the geographical area of origin. The diagnostic criteria Education Resource for Children and Adolescents (DER-CA). The DER-CA for the definition of GDM are those of IADPSG. Maternal Characteristics and database contains maternal diabetes status during pregnancy collected pregnancy outcomes were analyzed. through careful history and corroborating information and includes women Results: The average GDM prevalence was 4.2%, increasing since 2010 diagnosed with type 2 diabetes in childhood whose pregnancies were thus (2.6%) to 2015 (4.6). The prevalence is higher in foreign (6.3%) than the Ital- all complicated by PGDM. Linkage of mother-child dyads is possible within ian (3.5%) with a statistically significant difference (p <0.0001). Among the Therapeutics the Repository. ICD 9/10 diagnosis codes and physician tariff codes were foreigners, the prevalence is greater, with a statistically significant differ- Clinical Diabetes/

PUBLISHED ONLY used to identify diabetes in the biologic mother of children with type 2 dia- ence (p <0.0001) in African (9.5%) and Asian (10.5%) nationalities. betes identified from the DER-CA database. The timing of the diagnosis of In GDM mothers there is an excess of stillbirths, born with birth defects diabetes in the mother with respect to the gestational age of the pregnancy and an excess of preterm, low birth weight and hospitalized at birth (statisti- was determined. cally significant) infants. Results: The best administrative definition of GDM exposure was any inci- Conclusions: GDM is increasing in Trento province. Excess in foreign, par- dent code for diabetes (diabetes or GDM) > 25 weeks gestation and < 12 ticularly in some specific ethnic groups are confirmed, as well as an excess of weeks post-partum (sensitivity 81%, PPV 64%). PGDM exposure was best adverse events for newborns. Given the recent migration flows it has recom- defined in administrative data as an incident code for diabetes (excluding mended an effective state control model. GDM) in the mother prior to the pregnancy of the index child or within the first 24 weeks (inclusive) of pregnancy (sensitivity 91%, PPV 92%). 2480‑PUB Conclusion: We validated administrative definitions for GDM and PGDM Identifying Women with Recent Gestational Diabetes at Greatest exposure. The PGDM definition has superior performance. These definitions Need of Early Postpartum Testing for Diabetes can be used to differentiate GDM and PGDM exposure in studies utilizing MICHELE DREHMER, CRISTINA CASTILHOS, ADRIANA COSTA E FORTI, RUBEN administrative data. LADWIG, BRUCE B. DUNCAN, MARIA INÊS SCHMIDT, Porto Alegre, Brazil, Supported By: Children’s Hospital Research Institute of Manitoba Fortaleza, Brazil Background: Gestational diabetes (GDM) is a strong risk factor for future 2478‑PUB diabetes, which justifies postpartum screening. Yet, gives the overwhelming Glycated Haemoglobin in Gestational Diabetes Predicts Need for duties of recent motherhood, adherence to testing is low, particularly early Medical/Obstetric Intervention and Higher Postnatal Glucose Levels on. The purpose of this study is to assess diabetes risk and related factors MALIHA IQBAL, LING LING CHUAH, STONNY E. JOSEPH, Margate, United Kingdom, over the first 20 months postpartum among women with GDM participating Ashford, United Kingdom in the LINDA-Brasil Study. The value of glycated haemoglobin (A1c) in gestational diabetes (GDM) Methods: LINDA-Brasil, an ongoing clinical trial of diabetes prevention is controversial. Its routine use is not recommended despite its link to fetal postpartum, is nested within a cohort of women with recent GDM. The weight (FW) in established diabetes. The role of A1c in predicting the need cohort currently comprises 2856 women with GDM recruited from high risk for medical/obstetric intervention (I) and future maternal metabolic state is pregnancy clinics. We used various approaches to stimulate OGTTs after 8 not known. We set out to explore the relationship between maternal A1c, I weeks postpartum, including nearby, no cost testing, phone reminders and and 3 months post-natal glycaemic status (PNG). free clinic visits. Cox regression was used to estimate relative risks (RR) of Retrospective analysis of 25 GDM pregnancies was performed. Data developing diabetes. on maternal weight, A1c, metformin and/or insulin use, FW and PNG were Results: Mean (SD) BMI before pregnancy was 30.1 (6.4) kg/m2 and mean obtained. Patients were divided into 2 groups; 10 higher (HA) and 15 lower age 31.9 (6.2) y; 19.3% used insulin during pregnancy and an additional (LA) A1c using 35mmol/mmol (5.3%) as cut off. The pregnancy outcomes 16.3% oral hypoglycemic agents. By six months postpartum, 44% had under- were compared and unpaired test performed with p value< 0.05 deemed gone diabetes screening; among those who completed an OGTT, diabetes significant. Data is expressed as mean±SD. incidence was 2% for non-insulin users and 16% for insulin users. By 20 A1c was performed at a mean gestational age of 29.4±2.9 (HA) and months postpartum, 110 (8%) of those tested had developed diabetes, with 27.7±3.7 weeks (LA). BMI and blood pressure was higher in the HA group corresponding incidences being 4% and 27%, respectively. Use of insulin compared to LA (30.4±5.3 vs. 28.8±5.2kg/m2; 125.6±20.6/69.4±10.8 vs. and of hypoglycemic medication during pregnancy were associated with 112±8.3/65.1±9.3mmHg, p=0.035) as was fasting glucose (5.9±0.8 vs. RR of developing diabetes of 6.0 (95% CI 3.9-9.1) and of 1.9 (95% CI 1.3- 4.5±0.4mmol/L, p<0.002). 50% of the HA required Metformin/insulin and 2.9), respectively. Each 5 unit increase in BMI increased risk by 20% (RR 1.2; 40% required obstetric intervention (only 25% of LA required any form of I). 95% CI 1.0-1.4). Most incident cases used insulin (65%) or oral hypoglycemic PNG was also significantly higher in the HA group (e.g., 2 hour post glucose agents alone (17%) during pregnancy. load levels of 8.0±2.1 vs. 5.1±0.9mmol/L, p<0.002). There was a non-signifi- Conclusion: Use of insulin or oral hyperglycemic agents during pregnancy cant trend toward higher FW in HA. identifies most future cases of diabetes developed over 20 months postpar- Our data suggest that GDM patients with A1c above 35mmol/mol have a tum, constituting a simple means of targeting those at highest need of early worse metabolic/haemodynamic profile and are more likely to require inter- postpartum testing. vention. They also had higher PNG and a tendency to macrocosmic babies. Supported By: Brazilian National Counsel of Technological and Scientific Devel- We conclude that A1c remains a valuable biochemical parameter in GDM opment (563942/2010-0); Eli Lilly and Company

For author disclosure information, see page A751. ADA-Supported Research

A646 EPIDEMIOLOGY—AGINGCATEGORY

2481‑PUB Association between Peroxisome Proliferator-Activated Receptor Pro12Ala Polymorphism and Gestational Diabetes Mellitus DI MAO, PING LI, SHUO LIN, LING LI, JIANHUI FAN, Guangzhou, China The polymorphisms of peroxisome proliferator-activator receptor- gamma2 (PPAR-γ2) have been suggested to affect glucose metabolism. This study aims to explore the possible association between PPAR-γ2 gene Pro- 12Ala polymorphism and gestational diabetes mellitus (GDM). Two hundred pregnant women diagnosed as GDM and 200 non-GDM pregnant women in our hospital were recruited from June 2015 to December 2015. There were no statistically significant differences in baseline characteristics between the two groups. The Pro12Ala polymorphism (genotype Pro/Pro, Pro/Ala) were found in both GDM and control groups. A significant difference was observed in frequencies of Ala allele between GDM group (4%) and control 2484‑PUB group (8.8%) (P<0.05, odds ratio 0.49 (95% CI 0.26-0.92)). There were no A Comparison of Characteristics of Elderly and Younger Adults with statistical differences between genotype and prenatal BMI, fasting glucose, Type 2 Diabetes Mellitus (T2DM) after Failure of Oral Antidiabetic HDL, triglyceride, LDL (P>0.05). Our study suggested that PPAR-γ2 Pro12Ala Drugs (OAD) polymorphism were associated with GDM. Ala allele in PPAR-γ2 may be an JI HAI CHEN, XIAO JUN OUYANG, TIAN TIAN LU, KE WANG, YUN CHEN, LI SHEN, protect factor of GDM. RONG WEN BIAN, Nanjing, China, Shanghai, China The aim of this study was to compare the glycemic profiles of elderly 2482‑PUB vs. younger Chinese patients with T2DM at the time of insulin initiation to Assessing the Outcome of Universal vs. Risk-Based Screening for provide rationale for tailored insulin therapy. Data were extracted from a GDM in Resource Limited Setting of the Niger-Delta diabetes patient registry database of Jiangsu Province Geriatric Hospital, ROSEMARY N. OGU, OSITA C. JOHN, OMOSIVIE MADUKA, SUNDAY CHINENYE, Nanjing, China. Medical records of patients with T2DM (aged ≥18 years Port Harcourt, Nigeria [yrs] and HbA1c ≤7%) who had initiated insulin therapy from Jan 2005 to Diabetes is a growing NCD epidemic. In Pregnancy, early detection is Sep 2014 were analyzed. Data were stratified into 2 groups: elderly patients essential to improve fetomaternal outcomes and mitigate future type 2 dia- (≥65 yrs of age) and younger patients (<65 yrs of age). Patient demographics, betes and co-morbidities. disease duration, glycemic profiles, and OAD usage prior to insulin initiation Objective: To assess the outcome of universal or risk-based Screening for were compared between groups. A univariate analysis followed by multiple GDM in a resource-limited setting. logistic regression was applied to confirm the findings. Data from 165 Chi- Methods: A retrospective cohort survey of antenatal clinic attendees who nese patients with T2DM (elderly: 104; younger: 61) were analyzed. Demo- accessed care between January to October 2014 when screening for GDM graphic characteristics were similar in both groups, the majority of patients was universal and November 2014 to October 2015 when screening for GDM were male (61.5% vs. 75.4%, P=0.07). Duration of T2DM was higher among was selective risk-based. Data relating to diagnoses of GDM using the new elderly patients than younger patients (14 vs. 10 yrs; P=0.0002). Usage of WHO criteria, maternal characteristics such as age, parity, maternal and OADs was similar in both groups (P=0.93), with a majority of patients using Genetics fetal outcomes such as gestational age at delivery, mode of delivery, birth 2 OADs (57% vs. 60%, P=0.93). Both groups had a similar HbA1c level (8.54 Epidemiology/ weight, Apgar scores, etc were retrieved. Data analysis was done using vs. 8.77%; P=0.39). However, elderly patients had significantly lower fasting PUBLISHED ONLY SPSS version 21. (8.09 vs. 8.97mmol/L; P=0.045) and numerically higher postprandial glucose Results: Prevalence of GDM was 15.2% when universal screening was levels (15.1 vs. 13.7 mmol/L; P=0.12) prior to insulin initiation than younger done and 3.2% when selective risk-based screening was done. Multiple patients. The mean blood glucose [BG] excursion level (mmol/L) was sig- logistic regression analysis showed that for every unit increase in parity, nificantly higher among elderly patients as compared with younger patients women had a 63% greater odds of being screened for GDM at the antenatal (6.97 vs. 4.62; P=0.009), which was confirmed by the multiple logistic regres- clinic. (Odds ratio = 0.63; p value=0.00; C.I = 0.50 to 0.79). With selective risk- sion (OR [95% CI]: 1.17 [1.02, 1.35]; P=0.03). The present analysis showed based screening, there was no significant difference between Apgar scores that Chinese elderly patients with T2DM after OAD failure had significantly and the prevalence of stillbirths in babies born to women diagnosed with greater BG excursions than younger patients. Tailored therapy may need to GDM compared with babies born to women not screened for GDM. be chosen accordingly. Conclusion: Selective risk-based screening for GDM missed cases of GDM. The need for universal screening for GDM in resource-limited settings 2485‑PUB is reiterated. Real-Life Efficacy and Safety of Adding SGLT2 Inhibitors to Elderly DM2 Already Treated with DPP-IV Inhibitors or GLP-1R Agonists CARLOS TRESCOLI, PATRICIA PLATERO, JOSE ALEJANDRO ARAZO, EDUARDO EPIDEMIOLOGY—AGING ROVIRA, ANA TRESCOLI, ANDREA PEREZ, Alzira, Spain Management of DM2 in Elderly is challenging because little research, 2483‑PUB morbidities, frailty, polypharmacy and risk of hypoglycemia Combination of new antidiabetic drugs (ADD) claim to be more effective and safe. WITHDRAWN We evaluated the real world safety and efficacy of adding an SLGT2 Inhib- itor to an Elderly not controlled DM2 already treated with a DPP-4 Inh or a GLP-1 RA at least for one year and if their outcomes are different. We studied 102 Elderly DM2, treated with DPP-4 Inh (n= 54) and GLP-1 RA (n= 48) besides other ADD. SLGT-2 Inh. was added during a mean period of 19, 5 months in a Spanish Health Department. Data was collected from com- puterized Primary and Hospital Records before and after starting SLGT-2 Inh. Baseline Data: (DPP-4 Inh. and GLP-1 RA) Mean age 70,2 vs. 69,2 years, Mean period DM2: 12 vs. 13,4 years. Comorbidities: Hypertension 82 vs. 85%, Hypercholesterolemia 88 vs. 85%, BMI > 30 56 vs. 62% and had a previous cardiovascular or heart failure event: 50 vs. 53%. Baseline ADD: Metformin 76,5 vs. 82,3%, Insulin 32,3 vs. 38,2%, Sulphonylurea 38,2 vs. 26,5%, Repaglinide 20 vs. 9% and Pioglitazone, 9 vs. 15% Results pre- and post-SLGT-2 Inh: (DPP-4 Inh and GLP-1 RA) HbA1c: 8,8 vs. 7,6%, (S) and 8,4 vs. 7,3%, (S); Weight: 87,4 vs. 83,0 kg, (S) and 92,9 vs. 88,7 kg, (S); Systolic Blood Pressure (SBP): 134,4 vs. 133,1 mmHg, (NCS) and 136´7 vs. 131,2 mmHg, (S); There were no significant differences in Heart Rate, DBP, Lipid Profile, and Urinary function before and after SLGT-2 Inh

ADA-Supported Research For author disclosure information, see page A751.

A647