CLINICAL THERAPEUTICS/NEW TECHNOLOGY—GLUCOSECATEGORY MONITORING AND SENSING

2226-PUB CLINICAL THERAPEUTICS/NEW TECHNOLOGY— Development of the -Specifi c Risk-Taking Inventory GLUCOSE MONITORING AND SENSING RACHEL M. WASSERMAN, CAROLINE GONYNOR, BARBARA J. ANDERSON, DAVID D. SCHWARTZ, Houston, TX 2228-PUB Risky health behaviors are the primary cause of morbidity and mortal- Limitations of Blood Glucose Self-Monitoring in Patients with Dia- ity for all youth. For youth with type 1 diabetes (T1D), there are additional betes: A Baseline Analysis of the MyStarT Study opportunities to take risks with illness management, especially as many par- HELMUT ANDERTEN, GUIDO FRECKMANN, ANJA BORCK, BERNHARD KULZER, ents transition responsibility for diabetes care to the adolescent. While typi- Hildesheim, Germany, Ulm, Germany, Berlin, Germany, Bad Mergentheim, Germany cal risk-taking behaviors (e.g., alcohol and drug use) have been researched in Background: Patient-based monitoring of blood glucose (BG) has been shown youth with T1D, little is known about diabetes-specifi c risk behaviors. Thus, to improve diabetes control. While most devices are limited to the measure- we developed a diabetes-specifi c risk-taking inventory (DSRI) for adoles- ment of BG alone, MyStar Extra (Sanofi ) also calculates HbA1c estimates, a cents and young adults with T1D. marker commonly used to monitor glucose control over the longer term. It was The target population was youth with T1D (age 15-25). Initial items were the aim of the study to identify determinants of MyStar Extra use. generated based on clinical experience, theories of risk-taking, and review Methods: MyStarT is a non-interventional study conducted at offi ce of published scales of general risk-taking. Feedback on items was solicited based physicians across Germany. Adult patients were included, given that from 5 endocrinologists, 2 endocrinology nurse practitioners, and 4 psychol- they had newly received the MyStar Extra device at baseline. ogists with experience in diabetes research. Based on this input, items were Results: A total of 1,797 patients were included of which 1,521 patients refi ned, dropped, or added until consensus was reached. The questionnaire had a complete 24-week follow-up (85%). Patients had a mean age of was then piloted with 4 youth with T1D (ages 17-19) and cognitive debrief- 60.3±14.0 years, 56.1% were male, and the mean body weight was 90.8±20.2 ings were conducted to ensure participants’ understanding of questions’ kg. Diabetes type was type 1 and 2 in 13.8% and 86.0% of cases, respec- meaning and intent. tively and all received any form of treatment. At baseline, 42.2% of Thirty-six diabetes-specifi c risk-taking behaviors were identifi ed, includ- patients measured their pre-prandial blood glucose at each meal, 6.4% their ing 4 items for insulin pump users. A multiple-choice response format was post-prandial blood glucose at each meal and the mean HbA1c value was agreed upon to assess frequency of each behavior. All cognitive debriefi ng 7.7±1.9%. Overall 64.3% measured their blood glucose 7-times a participants noted the scale was generally easy to understand. Several items week. Reasons to opt for MyStar Extra were (multiple answers possible) a were changed to improve clarity based on cognitive debriefi ng feedback.

Therapeutics patient wish for a new device (38.0%), wish for added functionality (35.8%), The present fi ndings support the preliminary face validity of the inventory.

Clinical Diabetes/ outdated or defect device (28.5%), a need for closer monitoring (21.0%), and The DSRI may be a useful tool in assessing the prevalence of diabetes-spe- PUBLISHED ONLY others less frequent reasons. Only 9.7% of patients had no glucose meter cifi c risk-taking behaviors and perceptions of risk. Further research is needed available before study entry. to determine validity and reliability of the DSRI. Future research in this area Conclusions: The results illustrate that, even in patients with a device may guide clinical discussion and development of interventions to prevent available, the potential to optimize blood glucose control is not fully uti- adverse health outcomes. lized. Supported By: National Institute of Diabetes and Digestive and Kidney Diseases Supported By: Sanofi -Aventis Deutschland GmbH

2227-PUB 2229-PUB Patient and Educator Engagement in Developing a Connected System Linking Mobile-based Diet and Activity Self-Monitoring WITHDRAWN Data with Electronic Diabetes Education System for Behavioral Goal Monitoring JING WANG, LINDA M. SIMINERIO, CHIN-FUN CHU, DEIDRA COLEMAN, Hous- ton, TX, Pittsburgh, PA Chronicle Diabetes is a web-based system for ADA recognized educa- tion programs on diabetes education documentation. However, follow-up on patient behavioral goals has proven to be challenging for both patients and diabetes educators despite the availability of this tool. After rounds of inter- views seeking educators’ insights on their needs, we developed an interface in the Chronicle Diabetes system that enables educators to view patients’ self-monitoring data, captured via a fi tness tracker and its companion smart- phone application. This self-monitoring data was presented in Chronicle Dia- betes in a monthly calendar view, and included weekly totals and averages calories consumed and burned, steps, and macro-nutrient content. The aim of this study was to describe the patient and educator engagement during 2230-PUB the development and usability evaluation processes of this connected system. A think-aloud protocol was developed for user testing with 7 educa- WITHDRAWN tors to identify their concerns with the use of the connected system in their work setting and for follow-up on behavioral goals related to diet and physi- cal activity. After further refi nement on the system based on the think-aloud testing with the educators, a 2-week pilot study was conducted to test the connected interface with 8 patient-educator dyads. Patients and educators rated highly on perceived usefulness and acceptability of the system. Quali- tative interviews further showed the system in facilitating patients setting realistic goals, receiving more personalized care, and improving communica- tion with educators. Patient and educator experiences, issues, and sugges- tions in using the connected system were summarized. Engaging patients and educators in the development and testing of the connected system improved system usability, lending promise for easy adoption and dissemi- nation of the system in a larger scale. 2231-PUB Supported By: Robert Wood Johnson Foundation Blood Glucose Profi le in Chinese with Diabetes in Real World JIANZHONG XIAO, LING-WANG AN, Beijing, China Background: Self- (SBGM) is important for diabe- tes care. Data that describe the blood glucose profi le in Chinese with diabe- tes in real world were rare. Method: In collaboration with Tencent, from May, 1st, 2015 to October 31st 2015, patients with diabetes from 6 hospitals of Beijing Ruijing Diabe- tes Hospitals Limited Company received meters and test strips for free. The

For author disclosure information, see page A696. ADA-Supported Research

A564 CLINICAL THERAPEUTICS/NEWCATEGORY TECHNOLOGY—

SBGM results at home were automatically transferred to the Tencent Infor- years, 56.1% were male, the mean body weight was 90.8±20.2 kg, the mation Center. Five thousand and one hundred ninety four patients recruited mean HbA1c value was 7.7±1.9% and 13.8% had type 1 and 86.0% type (men= 2692, women=2502, aged 60.9±13.4). 2 diabetes. Mean patient satisfaction (available for n=1006) with their Result: A total 266,385 SBGM measurements were recorded. The fre- old device was 69.1 points and showed a mean increase of 10.4 (95% CI quency of SBGM was 12.9/month. 1.) Time of measurement: When 24 hours 8.6 to 12.3, p<0.0001) points 12 weeks after the introduction of the study divided into 4 six-hour equally, 51.2% data came from 5 AM to 11 AM, only device. The mean satisfaction with the study device indicated by the phy- 3.6% BG measurements came from 11 PM to 5 AM. 2.) : BG sician was 83.5 points (available for n=236). The correlation between phy- of 3901 measurements was lower than 3.9 mmol/l. The hypoglycemia pre- sician and patient satisfaction was 0.35 (95% CI 0.3 to 0.41; p<0.0001). The sented more frequently in the evening (10.13% of all tests form 1 AM to 2 AM). majority of patients (42.5% very likely and 44.5% probably) would recom- 3.) : 40% of post-breakfast and 42.4% of post-lunch BG were mend the study device. greater than 10 mmol/l. 4.) Fasting BG: The BG between 0 AM and 6 AM Conclusions: The use of the study device was associated with a high phy- was referred to fasting BG, proportion of patients reached the target goal sician satisfaction and an improved patient satisfaction. The device gets (3.9-6.9 mmol/l) for corresponding hour were 35.68%, 36.08%, 41.67%, recommendation from the majority of patients in this study. 45.49%, 42.56%, 41.30%, respectively. 5.) Time trend of BG: Among 711 Supported By: Sanofi -Aventis Deutschland GmbH cases monitored for 6 months, the frequency of hypoglycemia decreased from 1.89% to 0.98%. Patients with higher BG in the fi rst month got a better glycemic control with 6 months (from 0.5mmol/l to 4.8 mmol/l for BG>10 CLINICAL THERAPEUTICS/NEW TECHNOLOGY— mmol/l). INSULINS Conclusion: The majority of patients with diabetes did not reach target goal of blood glucose. Hypoglycemia was more common between 1 am to 2 am. SBMG is useful for improving diabetes control both for hypoglycemia 2234-PUB and hyperglycemia. Evidence of Real-World Effectiveness of Co-formulation of / in Insulin-Naïve Patients: An Initial Indian Supported By: Beijing Ruijing Diabetes Hospital Limited Company Experience RAJIV KOVIL, MAITREE LALAJI, NAVNEET WADHWA, Mumbai, India 2232-PUB Insulin degludec/insulin aspart (IDegAsp) (Novo Nordisk Ryzodeg® Accuracy and User Performance for the Dario™ Blood Glucose Mon-

70/30) is the fi rst soluble co-formulation combining a long-acting insulin Therapeutics itoring System (DBGMS)

degludec (IDeg) and rapid-acting insulin aspart (IAsp) commercially avail- Clinical Diabetes/

PAUL ROSMAN, BILLY PERY, SATISH K. GARG, New York, NY, Caesarea, Israel, able in India since January 2015. Recommended starting dose in insulin PUBLISHED ONLY Aurora, CO naïve patients with T2DM is 10 units once daily. We retrospectively evalu- The DBGMS enables self-monitoring of capillary blood (CB) glucose using ated the effectiveness in the real world setting with customised variable a smart phone and cloud based indexing. This study was performed to com- dosing approach as per the latest ADA guidelines (REALITY). The study pare the 1.) DBGMS glucose testing accuracy compared with YSI 2300 Stat participants (n=50, males =24, females=26) had mean age of 60.7 years Plus glucose analyzer results (Yellow Springs Instruments, Yellow Springs, ±SD 7.3. Baseline HbA1c was 9.7% ± SD 1.9 (range 6.2-14, 95% CI 9.1- Ohio), and 2.) the DBGMS usability by lay person operating the device for 10, p< 0.0001), duration of diabetes 13.5 years ± SD 5.1 (range 6-25, 95% the fi rst time, guided by the labeling material. 100 subjects (37 males or CI 12.1-15.02, p< 0.0001). At 3 months follow up the HbA1c reductions 63 females, with 93 type 1, 6 , 1 undetermined), ages 13 to mean±SD (-1.73±0.36) were statistically signifi cant (baseline 9.69±0.27 vs. 80 years were recruited at a single site. Meter, mobile device and strip lot at 3 months 7.96 ± 0.23; 95% CI -2.447 to -1.013; p< 0.0001). Post ther- numbers were recorded. Participants tested CB on the meter and within 5 apy decrease in HbA1c 7.9± SD1.6 (range 5-12.6, 95% CI 7.49-8.44; p< minutes, the professional assistant (PA) collected CB from a different fi nger 0.0001) and change in HbA1c -1.7± SD 1.7 (range -5.9- 1.9 CI -2.22 to -1.23 prick for the YSI and Hematocrit measurements. First time users completed p< 0.0001) were statistically signifi cant. IDegAsp in our real world set- questionnaires on DBGMS ease of use, while PAs documented their perfor- ting achieved greater absolute glycemic response in the higher baseline mance. The Table describes the accuracy of CB glucose by DBGMS using ISO HbA1c and higher baseline HbA1c level at initiation is a strong predictor 2013 criteria and ease of use results. We conclude that the Dario™ BGMS is of HbA1c reduction. Future research on larger, comparative real world set- accurate and user friendly. ting is warranted. Table. Table. Reduction in HbA1c Compared across the Age Groups, Duration of Diabetes, and Baseline HbA1c. Demographics Mean ± SD 95% Confi dence Interval p value Age (years) <60 (n=22) -1.9 ± 0.34 -0.62 to 1.41 =0.43 (NS) ≥ 60 (n=27) -1.5 ± 0.36 Duration of Diabetes (years) ɖ 10 (n=17) -2.17 ± 0.38 -0.37 to 1.7 = 0.57 (NS) > 10 (n=33) -1.5 ± 0.31 Baseline HbA1c (%) ɖ 9 (n=22) 0.75± 0.24 -2.6 to -0.85 = 0.0002 (Signifi cant) > 9 (n=28) -2.4 ± 0.33

2233-PUB Patient and Physician Satisfaction Using the MyStar Extra Device, 2235-PUB a SMBG Meter that Provides Estimated HbA1c Values Therapeutic Regimen of 3-time Premixed Insulin for Brittle Diabetes: GUIDO FRECKMANN, BERNHARD KULZER, ANJA BORCK, HELMUT ANDERTEN, A Self-Controlled Study Ulm, Germany, Bad Mergentheim, Germany, Berlin, Germany, Hildesheim, Germany LU LIU, PENG YANG, YILI LUO, MING CHEN, LU XU, LE BU, MANNA ZHANG, ZIWEI Background: Self-monitoring of blood glucose (SMBG) plays an important LIN, FENG LI, HONG LI, SHEN QU, Shanghai, China role in treatment control and adaptation of people with diabetes. Recently Brittle diabetes (BD) is a frustrating challenge for diabetes treatment in a SMBG, the MyStar Extra device (Sanofi ), was introduced that estimates clinic. Insulin pumps and basal insulin with three fasting insulin injections HbA1c in addition to glucose values. In this study the satisfaction of patients are the most effective treatment. However, these two methods have the and their physicians with the use of this device was evaluated. shortcomings of high cost, inconvenience and poor compliance. Therefore, Methods: The non-interventional MyStarT study was conducted at offi ce in this study, three-time premixed insulin (3-TPI), but not analogues in one based physicians across Germany. Adult patients were included, given that insulin pen method was assessed in BD patients. Seventy-three BD patients they had newly received the MyStar Extra device. Satisfaction with the use of (including 36 male and 37 female, 38 T1DM and 35 T2DM) were enrolled and the device was indicated by both physicians and patients using a visual ana- divided into three groups according to their previous therapeutic regimen: logue scale graded “not satisfi ed at all” [0] to “very satisfi ed” [100)]. Patients two-time premixed insulin group (2-TPI, n=23), three-time completed the questionnaire at baseline and the 12-week follow-up. with one time neutral protamine hagedorn insulin group (3R+N, n=30), three- Results: A total of 1,797 patients were included of which 1,521 patients time regular insulin with one time group (3R+G, n=20). All had a complete follow-up (85%). Patients had a mean age of 60.3±14.0 patients have volunteered to receive 3-TPI therapy regimen for six months

ADA-Supported Research For author disclosure information, see page A696.

A565 CLINICAL THERAPEUTICS/NEWCATEGORY TECHNOLOGY—INSULINS

during this study, and then the HbA1c level, incidence rate of hypoglycemia, (DET), NPH, degludec (DEG), glargine biosimilar (BIO) and premixed insulin dosage of insulin, body weight and therapy compliance were measured. The (PRE). A total of 53 RCTs were analyzed. Change in HbA1c was comparable results indicated that the HbA1c level (2-TPI: 11.25 ± 1.32% vs. 8.69 ± 0.30%; between Gla-300 and Gla-100, DET, NPH, DEG, BIO and PRE. Gla-300 was P=0.016; 3R+N: 11.06 ± 1.05% vs. 8.74 ± 0.43%; P=0.023; 3R+G: 8.95 ± associated with a signifi cantly lower documented symptomatic hypoglyce- 0.56% vs. 8.63 ± 0.30%, P=0.054), incidence rate of hypoglycemia (times per mia (HYPO) event rate vs. Gla-100 (relative risk: 0.77; 95% confi dence inter- month) (2-TPI: 9.8 ± 3.4 vs. 2.1 ± 0.8 P=0.041; 3R+N: 8.4 ± 3.0 vs. 2.2 ± 1.0, val: 0.72 to 0.81), DET (0.67; 0.6 to 0.73), NPH (0.55; 0.5 to 0.6), DEG (0.49; P=0.027; 3R+G: 5.1 ± 2.5 vs. 2.1 ± 1.2, P=0.038) and dosage of insulin (2-TPI: 0.44 to 0.54), BIO (0.76; 0.7 to 0.84) or PRE (0.51; 0.47 to 0.54). For severe 0.50 ± 0.09 vs. 0.46 ± 0.08 U/kg·d; P=0.048; 3R+N: 0.67 ± 0.10 vs. 0.52 ± HYPO event rate, Gla-300 was associated with a signifi cantly lower rate vs. 0.07 U/kg·d; P=0.039; 3R+G: 0.58 ± 0.09 vs. 0.50 ± 0.09 U/kg·d, P=0.072;) of NPH (relative risk: 0.38; 95% credible interval: 0.14 to 0.94), BIO (0.14; 0.02 most groups have signifi cantly decreased after the treatment. Additionally, to 0.75) and PRE (0.21; 0.08 to 0.54), with non-statistically signifi cant differ- 63% of the patients were satisfi ed with this therapy regimen, 70% patients ences when compared to DEG, DET or Gla-100. For body weight change from considered it convenient, and 88% patients think that the treatment cost baseline, Gla-300 is associated with lower weight gain compared to PRE was acceptable. In summary, this study indicated that 3-TPI in one insulin (-1.82kg; 95% credible interval: -2.66 to -0.98), higher vs. DET (0.79kg; 0.01 to pen therapy regimen was effective, convenient and economical, and it is an 1.67) and comparable with the other comparators. Sensitivity analyses con- alternative solution for the treatment of BD patients with poor compliance sidering subsets of studies, different HYPO defi nitions, and adjustment for and economic condition. trial-level characteristics confi rmed the robustness of these fi ndings. This NMA suggests that Gla-300 is associated with comparable glycemic control 2236-PUB to other basal insulin comparators, a signifi cantly lower risk of documented Reduced Insulin Requirements and Weight in Veterans with Type 2 symptomatic hypoglycemia when compared with most of basal insulin ther- Diabetes on U-500 Insulin and apies and less severe hypoglycemia events vs. NPH, BIO and PRE. PADMAJA AKKIREDDY, VIJAY SHIVASWAMY, CYRUS V. DESOUZA, Omaha, NE A signifi cant number of veterans with type 2 diabetes have severe insu- 2238-PUB lin resistance. Studies have shown that in patients requiring large doses of Meta-analysis of the Insulin Dosage in Chinese Type 2 Diabetes insulin, conversion to U-500 resulted in improvement of glycemic control, Patients Receiving Insulin Treatment along with undesirable weight gain and signifi cant increase in total daily XIAOLING CAI, WENJIA YANG, XUEYING GAO, LINGLI ZHOU, XUEYAO HAN, dose of insulin (TDD). We studied the effect of adding liraglutide on weight, LINONG JI, Beijing, China Therapeutics A1c reduction and insulin requirements in patients treated with U-500 insu- With the widely used insulin treatment in Chinese type 2 diabetes Clinical Diabetes/

PUBLISHED ONLY lin. We conducted a small open-labeled prospective study of 6 veterans with patients, the aim of this study is to evaluate the insulin dosage in Chinese type 2 diabetes on U- 500 treated with liraglutide titrated up to 1.8mg daily. type 2 diabetes patients receiving various kinds of insulin treatment. The All insulin dosage is expressed as U-100 equivalent. The average time on MEDLINE, EMBASE, Chinese National Knowledge Infrastructure (CNKI) liraglutide was 2.5 months (range- 1 to 5 months). The mean TDD of insu- and China WanFang databases were searched and qualifi ed studies were lin for these 6 patients at the start of the study was 400 ± 265 units (range included. The inclusion criteria were as following: 1.) randomized controlled 225 to 925 units). The mean reduction in weight was 2.8 ± 1.26 kgs (p<0.05) trial in type 2 diabetes patients; 2.) Chinese participants; 3.) insulin treat- (Figure 1). The reduction in insulin dosage was 117± 68 TDD of insulin (p<0.01) ment in one arm or both arms in the trial; 4.) study duration more than 12 (Figure 1). HbA1c decreased by 0.9 ± 0.5, from 8.9 ± 0.8 to 8.0 ± 0.5 (p<0.05). weeks. Totally 88 qualifi ed studies were included. According to randomized Interestingly the reduction in insulin dosage was disproportionately large controlled trials, in patients receiving basal bolus insulin treatment, daily compared to the actual weight loss. We conclude that, in the only study of insulin dosage for the long-acting in combination with rapid liraglutide effects in highly insulin resistant veteran population, addition of insulin analog was 28.30u/day, daily insulin dosage for the NPH combined liraglutide to patients on U-500 has benefi cial effect of decreasing the insu- with regular insulin was 42.16u/day, daily insulin dosage for the NPH com- lin requirements in addition to its weight loss effects. bined with rapid insulin analog was 39.41u/day. In patients receiving basal Figure 1. insulin with oral hypoglycemic agents treatment, daily insulin dosage for the long-acting insulin analog was 18.33u/day, daily insulin dosage for NPH insu- lin was 16.34u/day. In patients receiving premixed insulin treatment, daily insulin dosage for premixed insulin analog with oral hypoglycemic agents was 26.41u/day, daily insulin dosage for premixed insulin analog alone was 36.14u/day, daily insulin dosage for premixed human insulin with oral hypo- glycemic agents was 29.91u/day, daily insulin dosage for premixed human insulin alone was 38.42u/day. This meta-analysis was the fi rst meta-analysis focused on the insulin dosage in Chinese type 2 diabetes patients which provided more comprehensive clinical evidence on the insulin dosage among different insulin treatments in treating Chinese type 2 diabetes patients.

2239-PUB The Combination of CJC-1134-PC and CJC-1575 Has Potential to Improve Glycemic Control while Attenuating Weight Gain XIPING LIU, SEAN HONG, SHIJUAN WU, ZHIWEN YAO, QIANG X. YAO, Culver City, CA The combination therapy of basal insulin with a GLP-1 receptor agonist (RA) offers a promising approach for the treatment of diabetic patients because of the synergistic effects on improved glycemic control and the complementary effects on stable body weight. Several clinical trials that 2237-PUB have studied the addition of RA to ongoing insulin, or addition of insulin to Safety and Effi cacy of Insulin Glargine 300 u/mL (Gla-300) Compared RA showed improved glycemic control while lowering the risk of hypoglyce- with Other Basal or Premixed Insulin Therapies in Patients with mia and weight gain. However, most of the available combination regimens Type 2 Diabetes Mellitus (T2DM): A Network Meta-analysis (NMA) require frequent dosing, which is inconvenient to patients. We have devel- NICK FREEMANTLE, HONGWEI WANG, DEEPIKA THAKUR, PETER STELLA, NEHA oped basal insulin, CJC-1575, and once-a-week exendin-4 analogue, CJC- ™ CHANAN, MANIK KALRA, JASSO-MOSQUEDA JUAN GUILLERMO, London, 1134-PC, using our proprietary Drug Affi nity Complex (DAC ) technology. United Kingdom, Bridgewater, NJ, Waltham, MA, Chilly Mazarin, France Currently, the combination regimen of these two compounds is being inves- Gla-300 is a new insulin recently approved in the U.S. and EU with a fl at- tigated. In vitro functional studies showed that CJC-1134-PC does not inter- ter more prolonged PK/PD profi le compared with glargine 100u/mL (Gla-100). act with CJC-1575 with respect to cAMP response and CJC-1575 does not The aim of this study was to assess the relative effi cacy and safety of Gla- interact with CJC-1134-PC with respect to phosphorylation of insulin recep- 300 in naive or already exposed basal insulin T2DM patients. Data from a tor (IR). Using a rat medullary thyroid carcinoma cell line (rMTC 6-23), which systematic literature review including randomized evidence between 1980 expressed both GLP-1R and IR, the phosphorylation of Akt S473 (pAkt) had and December 2014 were included. The comparators were Gla-100, detemir a dose dependent increase after treatment with CJC-1575. However, it was

For author disclosure information, see page A696. ADA-Supported Research

A566 CLINICAL THERAPEUTICS/NEWCATEGORY TECHNOLOGY—INSULINS found that the pAkt decreased after treatment with CJC-1134-PC. Interest- with 1.) intensive insulin therapy (4 injections daily) or 2.) an insulin pump, ingly, the decreased level of pAkt was offset in the cells treated with both and why. For both treatment scenarios, we explained the need to follow a CJC-1575 and CJC-1134-PC simultaneously. It has been reported that the diabetic lifestyle, check sugars 4 times a day, and communicate daily with suppression in pAkt mediates the food intake-suppressive effects of GLP-1R their doctor. We further explained that after 1 year, the chance of needing activation, which manifests clinically as nausea and loss of appetite. It is to take diabetes pills would be less than half (40%). Likelihood was reported plausible that this suppression of pAkt by CJC-1134-PC will be counteracted on a scale of 1 (not at all likely) to 10 (very likely). Thematic analysis was by the pAkt increase with CJC-1575. This data reveals insights of the combi- conducted with open-ended responses. We found that 57% of participants natorial mechanism of action of CJC-1134-PC with CJC-1575, leading way to were willing (≥7/10) to undertake a very intensive insulin treatment plan. favorable outcomes of limited risk of weight gain and gastrointestinal side Willing participants were equally divided between receiving either injec- effects while improving glycemic control. tions or the pump (n=11), only injections (n=12), and only the pump (n=11). Prominent themes included patients expressing preferences about insulin 2240-PUB and regarding the mode of insulin therapy. In general, participants willing Inpatient Insulin Requirements in Highly Insulin-Resistant Veterans to undertake intensive insulin therapy thought that 2 weeks was reasonably with Diabetes short and looked forward to the future possibility of not needing diabetes ROHIT R. KEDIA, CYRUS V. DESOUZA, VIJAY SHIVASWAMY, Omaha, NE . Unwilling participants considered the treatment plan too much Veterans with type 2 diabetes who are highly insulin resistant (> 200 work. This study suggests that patients with type 2 diabetes may be willing units/day) often require U-500 insulin (fi ve times more concentrated regular to undertake very intensive diabetes treatments for short periods of time in insulin) for better glycemic control. The insulin requirements in hospital of exchange for avoiding future medicines. These data are important for future such veterans that are using ambulatory U-500 insulin is not well studied. directions of research on novel approaches to diabetes management. We hypothesized that patients who were on U-500 insulin would require Supported By: National Institutes of Health signifi cantly less total daily insulin during hospital admission, and eventu- ally return to pre-admission doses at follow up. We performed a retrospec- 2242-PUB tive study of 26 veterans that were taking ambulatory U-500 insulin and A Pilot Bioequivalence (BE) Study of Ultra-Rapid Concentrated Bio- admitted to the hospital for a minimum of 24 hours and that were eating Chaperone Lispro (BCLIS) U200 to BCLIS U100 meals during admission. All patients were transitioned to U-100 insulin upon GRIT ANDERSEN, GRÉGORY MEIFFREN, BERTRAND ALLUIS, AYMERIC RANSON, hospital admission. Insulin dose changes over time were analyzed using RÉMI SOULA, MARTIN GAUDIER, OLIVIER SOULA, TIM HEISE, SIMON BRUCE, repeated measures ANOVA model. All insulin doses are expressed as U-100 Neuss, Germany, Lyon, France Therapeutics Clinical Diabetes/

equivalent. The total daily dose of insulin (mean±SD) during hospital stay BCLIS U100 is a 100 U/mL ultra-rapid formulation designed PUBLISHED ONLY (135±60 units) was signifi cantly less than prior to admission (288 ± 111 units) to better mimic the physiological timing of prandial insulin action. A con- (p< 0.0005) and at six week follow-up (300±99 units) (p<0.0005) (Figure 1). centrated 200 U/mL BCLIS formulation is developed to reduce injection vol- Veterans required less than 67% of their prior-admission dose during hos- umes for diabetic patients requiring high insulin doses. In a pilot double-blind pitalization. Given the potential for hypoglycemia, U-500 insulin should be replicated cross-over BE trial, 26 healthy subjects (mean±SD age 44.6±10.0 used with caution in hospitalized patients. Prospective trials are needed to yrs; BMI 24.5±2.2 kg/m²) received two subcutaneous 0.2U/kg doses of design algorithms to transition ambulatory patients on U-500 to U-100 insu- BCLIS U100 and BCLIS U200. Lispro concentration-time profi les were similar lin while hospitalized. after dosing with each concentration (Figure). Point estimate (Least Square Figure 1. Total Daily Insulin Required. Mean ratio) and 90% confi dence interval of BE parameters were within the [0.8;1.25] interval (Table). To conclude, the BE of BCLIS U200 to BCLIS U100 is shown in this pilot trial; ultra-rapid profi le of BCLIS is retained at 200 U/mL. Table. Statistical Analysis of the Pharmacokinetic Parameters Defi ning Bioequivalence. Geometric Point 90% Confi dence Interval Least Square Means Estimate for Point Estimate Parameter U200 U100 U200:U100 (Lower;Upper) Cmax (pg/mL) 3126 3526 0.89 (0.83;0.95) AUC-LIS 0-inf (h*pg/mL) 8712 8493 1.03 (0.99;1.06) AUC-LIS 0-t (h*pg/mL) 8566 8370 1.02 (0.99;1.06) AUC-LIS 0-1h (h*pg/mL) 2126 2461 0.86 (0.81;0.93) Early t [0.5 (Cmax)] (h) 0.248 0.240 1.04 (0.95;1.13) t1/2 (h) 1.002 0.965 1.038 (0.94;1.15)

Figure. Free Immunoreactive Insulin Lispro Concentration vs. Time Profi les.

2241-PUB Patient Willingness to Undertake 2-Weeks of Very Intensive Type 2 Diabetes Treatment for the Possibility of Cure NEDA LAITEERAPONG, AVIVA G. NATHAN, MICHAEL T. QUINN, ELBERT S. HUANG, Chicago, IL Some diabetes experts have advocated the use of early insulin therapy after diagnosis of type 2 diabetes in order to preserve beta-cell function and reduce the need for future medicines. However, research also suggests that patients are very reluctant to start insulin therapy. We sought to understand how willing patients may be to undertake insulin therapy for a short dura- tion, as a tradeoff for possibly avoiding future diabetes medications. We interviewed 60 patients with type 2 diabetes (<10 years), taking oral medi- cations, and asked how likely they would be to adopt a 2-week treatment

ADA-Supported Research For author disclosure information, see page A696.

A567 CLINICAL THERAPEUTICS/NEWCATEGORY TECHNOLOGY—INSULINS

2243-PUB Table. The Infl uence of a New Long-Acting Insulin on Type 2 Diabetes Mel- litus Regulation and Quality of Life in Low-Income Patients ANASTASIOS G. KOUTSOVASILIS, ALEXIOS SOTIROPOULOS, OURANIA APOS- TOLOU, ANNA GROZOU, PARASKEVI VERGIDOU, ANASTASIA KATOPODI, ISIDOROS BINIKOS, ILIAS TAMVAKOS, STAVROS BOUSBOULAS, Nikaia, Greece A new long-acting, but more expensive insulin, degludec, shows greater duration compared to already existing long-acting insulins. The aim of the study is to examine the effect of degludec insulin on type 2 diabetes mellitus (T2DM) regulation and quality of life in low-income patients since the nega- tive side of it is the high cost. 98 T2DM patients were enrolled in the study and received degludec insu- lin along with tablets or/and GLP-1 agonist or they were under a regimen of multiple injection. Six months later, their glycemic regulation, the units, the quality of life (SF-12), their compliance with the therapeutic regimen (using Morisky scale) as well as the fl exibility in insulin use were assessed. Out of the study’s 98 patients, aged 63.45 ± 10.88 years and diabetes duration of 10.52±6.48 years, 72.5% received degludec and tablets, 9.31% received GLP-1 agonist and 18.19% were under a regimen of multiple injec- tions, with the degludec-GLP-1 group having a higher BMI (p=0.029). All three groups show a signifi cant quality of life. became less frequent after the initiation of degludec (p=0.089). There was a signifi cant improvement in quality of life for all patients (p=0.044). A reduction in deglu- dec units was observed (16%) while the greater reduction was in the deglu- dec-GLP-1 group (22.8%, p=0.040). It was observed that 60.2% received insulin up to 8 hours after prescribed time at least 3 times per month and Therapeutics compliance also improved (p=0.041). Clinical Diabetes/ The use of degludec is associated with a lower rate of hypoglycemias, a PUBLISHED ONLY good quality of life and signifi cant compliance with the therapeutic regimen while many patients are utilize the possibility of fl exible use of deglude’s Supported By: Eli Lilly and Company; Boehringer Ingelheim insulin. It seems that degludec’s negative point is only the cost. 2245-PUB 2244-PUB Insulin Promotion: A Study of Insulin Advertising in Medical Journals Characteristics of Patients with Type 2 Diabetes (T2D) Who Suc- RYAN P. KNOX, VERONIKA J. WIRTZ, RICHARD O. LAING, MARG EWEN, Boston, cessfully Achieved A1c <7% MA, Amsterdam, Netherlands LIZA L. ILAG, DACHUANG CAO, MICHAEL CUMMINGS, IRENE HADJIYIANNI, Medical journals are a leading source of information on medicines for phy- ROBYN K. POLLOM, TIMOTHY COSTIGAN, MENG TAN, Indianapolis, IN, Ports- sicians, yet low quality journal advertisements are a global problem. There mouth, United Kingdom, Bad Homburg, Germany, Ann Arbor, MI are few studies analyzing advertisements in medical journals on a global Identifying factors related to achieving glycemic targets is important for scale, and none identifi ed specifi cally analyzed insulin promotion. The objec- improving insulin treatment effectiveness. Post hoc analyses of ELEMENT-2 tives of this study were to identify and collect insulin advertisements from data assessed characteristics of patients (pts) with T2D receiving insulin general and specialized medical journals, to describe the presence of insu- glargine U-100 who achieved A1c <7% or not at week (wk) 24. Of the 660 pts lin product advertisements in medical journals, to analyze their compliance on insulin glargine U-100 (pooled LY2963016 and Lantus® data), 352 (53%) with international recommended standards, and to analyze the claims, side achieved A1c <7% at wk 24; 308 (47%) did not, with differences (p<.05) noted effects, and warnings included in the advertisements. Insulin advertise- in baseline A1c, diabetes duration, prestudy insulin use, and use ments published in the 2005, 2010, and 2014-2015 issues of a convenience (Table). The A1c <7% group was more likely (p<.05) to achieve fasting blood sample of 27 general and specialized medical journals were collected from glucose (FBG) ɖ100 mg/dL and postprandial glucose (PPG) ɖ180 mg/dL (both 16 high- and middle-income countries. Images of the advertisements were based on self-monitored blood glucose), had greater (p<.05) decreases in analyzed based on the criteria suggested by the World Health Organization A1c, FBG, and PPG than the A1c ≥7% group, and similar (p≥.05) insulin dose and Health Action International. Content analyses were performed to iden- increase, hypoglycemia incidence rate, and hypoglycemia event rate (events/ tify the nature of the claims, side effects, and warnings included. Of the 288 pt/year) as the A1c ≥7% group. In summary, pts more likely to achieve A1c advertisements studied, all but three were published by one of the major <7% were insulin naïve, not on sulfonylureas, had lower baseline A1c, or had insulin manufacturers (Novo Nordisk, Sanofi , Eli Lilly). The overwhelming shorter diabetes duration. Pts with A1c ≥7% had less improvement in FBG majority (96%; n=277) were for analog insulin products. Prescribing informa- and PPG than the A1c <7% group while on a similar insulin dose. Thus, insulin tion (including side effects, major interactions, and warnings) was the sec- glargine U-100 is an effective starter basal insulin; suboptimal glycemic con- tion with the lowest compliance to international criteria. Compliance varied trol may be addressed by appropriate insulin/sulfonylurea dose titration or across the countries, although the sample was too small to show a statistical considering PPG-lowering treatment to allow dose uptitration while avoid- difference. As promotion is an important factor infl uencing the prescribing ing hypoglycemia. and inappropriate use of medicines, it is important to continuously monitor advertisements’ compliance with international standards. Less compliance with information on side effects and major interaction can result in the per- ception of infl ated product benefi ts. Supported By: Boston University

For author disclosure information, see page A696. ADA-Supported Research

A568 CLINICAL THERAPEUTICS/NEW TECHNOLOGY—INSULINCATEGORY DELIVERY SYSTEMS

2246-PUB Figure 1. HbA1c from Baseline and Mean Yearly HbA1c Throughout the Postoperative Glycemic Control after Distal Pancreatectomy Follow-up. AIQUN LIU, KIM A. CARMICHAEL, MARILYN E. SCHALLOM, W. DEAN KLINKEN- BERG, St. Louis, MO Distal pancreatectomy (DP) involves resection of lesions in the body and the tail of the . DP may lead to both insulin and defi - ciency, which may worsen diabetes or bring about pancreatogenic diabetes. Maintaining glycemic control can be challenging after DP. The purpose of this study was to assess postoperative glycemic control and management among patients with DP. Medical records from 82 eligible adult patients between 2013 and 2014 with DP were reviewed retrospectively. Twenty-one (25.6%) patients had pre-existing diabetes. The average length of hospital stay was 5.8 ± 2.6 days. The average volume of resected pancreas was 193 ± 313 cm3. A total of 2,124 blood glucose (BG) tests was performed, bedside or laboratory. Only 6 (0.3%) of the BG tests were <70 mg/dL (3.9 mmol/L), and 956 (45%) were >140 mg/dL (7.8 mmol/L). Among 82 patients, 75 (91.5%) had at least 1 BG level >140 mg/dL. The average patient had 11.7 BG tests >140 mg/dL (SD=12.7). Mean number of BG events >140 mg/dL per day was 1.71 ± 2250-PUB 1.61. Postoperatively, insulin was the most common agent prescribed for gly- Minimal Effective Doses in Continuous Subcutaneous Insulin Infu- cemic control. Among 68 patients receiving insulin, 86% used rapid-acting sion correctional insulin, 7% long-acting basal insulin, 4% meal-dose insulin, and ENRIQUE CAMPOS-NANEZ, HOWARD C. ZISSER, Charlottesville, VA, Billerica, MA 4% both insulin and oral agents. On postoperative day 1 through 6 and on the The goal of this work is to identify the minimum insulin dose that will day before discharge, less than 30% of patients received insulin. Of those result in a target glucose concentration change during basal and bolus deliv- receiving insulin, 50% or greater received <0.10 units/kg. Nineteen (23.2%) ery phases. patients were discharged on diabetic medications. At time of discharge, 3 We used the UVA/PADOVA T1DM Simulator to quantify these effects had home diabetic discontinued, 3 had medication reduced, 7 during continuous subcutaneous insulin infusion (CSII). During the entire had medication resumed, 5 were started on insulin and 1 had an oral agent simulation subjects are administered insulin basal rates to maintain their Therapeutics changed. Although hypoglycemia was minimal, when determining the appro- Clinical Diabetes/ glucose concentrations at 120 mg/dL (6.7 mmol/L). The following 2 changes PUBLISHED ONLY priate dose of insulin for patients with DP, clinicians may need to weigh the were simulated independently: 1.) insulin infusion rates are increased 1 hour risk of hypoglycemia against the benefi ts of glycemic control. Routine moni- into the simulation, and 2.) a single bolus is injected 1 hour into the simula- toring of point-of-care BG is warranted. tion. Both scenarios were simulated varying basal rate and bolus sizes in increments of 0.01 U/h and 0.01 U, respectively, for each subject in the child, 2247-PUB adolescent, and adult cohorts of the simulator. Our results show that in order to produce deviations of 25 mg/dL (1.4 WITHDRAWN mmol/L) from the basal state of 120 mg/dL (6.7 mmol/L) in 5% of the chil- dren, adolescent, and adult populations mentioned above, it is necessary to modify their basal rates by 0.06, 0.21, and 0.19 U/h, respectively. In scenario 2, a single bolus of size 0.19, 0.55, and 0.6 U, respectively, produced similar CLINICAL THERAPEUTICS/NEW TECHNOLOGY— deviations in 5% of the same populations. Additional results are reported in INSULIN DELIVERY SYSTEMS Table 1. Furthermore, we conclude that only 2% of the entire population, cor- responding to children with low body weight and high insulin sensitivities, will manifest the effect of insulin dose changes of 0.03 U/h. 2248-PUB Table 1. Changes to Insulin Infusion Required to Affect 5% of the Population. WITHDRAWN Basal Rate Change (U/h) Bolus (U) +/-12.5 mg/dL +/-25 mg/dL +/- 12.5 mg/dL +/- 25 mg/dL (0.7 mmol/L) (1.4 mmol/L) (0.7 mmol/L) (1.4 mmol/L) 2249-PUB Children 0.03 0.06 0.10 0.19 Insulin Pumps Are Effective and Durable in Adult-Onset Type 1 Dia- Adolescent 0.11 0.21 0.28 0.55 betes Adult 0.10 0.19 0.30 0.60 VINCENZO PROVENZANO, DAVIDE BRANCATO, ALESSANDRO SCORSONE, MATTIA FLERES, VITO AIELLO, MICHELE DE SANTIS, ANNA DI NOTO, ALES- SANDRO FERALDI, FRANCESCA PROVENZANO, GABRIELLA SAURA, LUCIA SPANO, Palermo, Italy 2251-PUB Continuous subcutaneous insulin infusion (CSII) is considered as a valu- Frequency of Use and Effi cacy of Continuous Subcutaneous Insulin able therapeutic option for people with diabetes, but there is a substantial Infusion (CSII) Therapy at a Referral Hospital in Spain lack of studies on long-term effectiveness of CSII in type 1 diabetes (T1D) PILAR I. BEATO-VÍBORA, FRANCISCO J. ARROYO-DÍEZ, LAURA GALÁN-BUENO, with adult onset. To assess the effectiveness and durability of CSII in people BEATRIZ GALVÁN-DÍAZ, CLAUDIA GARCÍA-LOBATO, ANA LÓPEZ-NAVIA, ESTHER with adult onset T1D, we studied 70 people treated with CSII (male/female = DELGADO-GARCÍA, Badajoz, Spain 35/35) and with age at onset > 35 years (age at onset = 44.7 ± 8.8 years Introduction: The frequency of use of CSII therapy in DM1 in Spain is lower [mean ± DS]; age at CSII commencement = 49.1 ± 10.4 years). Main outcome than the European average (15%). Extremadura (southwestern Spain) is one was the mean of glycosylated hemoglobin (HbA1c) values obtained through- of the regions in Spain with the highest CSII use. out the follow-up (MHbA1c). Secondary outcomes were the the yearly mean Aims: To record data from all DM1 patients being followed at Badajoz of HbA1c values (mHbA1c). After a follow-up of 1-8 years, MHbA1c resulted Hospital, a referral hospital in Extremadura, the modality of insulin treat- signifi cantly decreased from baseline HbA1c (respectively, 7.71% ± 1.08 SDS ment and their glycaemic control. vs. 8.21% ± 1.20; p < 0.001); the latter was directly related to MHbA1c and Methods: Data from the DM1 patients being followed at the Adult and was the only predictor of MHbA1c (R2 = 0.427). HbA1c decreased signifi - Paediatric Departments were registered. HbA1c before CSII and last HbA1c cantly from a baseline value of 8.2% ± 1.2 to mHbA1c values of 7.8% ± 1.2 were documented. (1st year), 7.6% ± 1.3 (2nd year), 7.6% ± 1.4 (3rd year), 7.7% ± 1.2 (4th year) Results: 779 patients were included (Table 1). Last HbA1c was signifi - (ANOVA for repeated measures: p < 0.05; F = 2.70), with a non signifi cant cantly lower in the CSII group than in the MDI group (7.2±0.8% vs. 7.6±1.2%, reduction from the 5th year (Figure 1). In conclusion, our study suggests that p<0.001, duration of diabetes 16.9±14.0 years vs. 17.2±11.5 years, p=0.8; CSII in T1D with adult onset is: 1.) effective to decrease HbA1c throughout a adults: 7.3±0.9% vs. 7.7±1.2%, p<0.001; children: 7.2±0.7% vs. 7.7±1.3%, prolonged follow-up, with a signifi cant trend for 4 years; 2.) more effective p=0.001). Percentage of patients with HbA1c ɖ7% was 46% in the CSII group when baseline HbA1c is lower. (50% in children and 44% in adults) and 30% in the MDI group (children

ADA-Supported Research For author disclosure information, see page A696.

A569 CLINICAL THERAPEUTICS/NEW TECHNOLOGY—NON-INSULINCATEGORY INJECTABLES

and adults); all p<0.05. In the CSII group, last HbA1c was lower than HbA1c 7 mmHg; p=0,000); dBP (82, 2 vs. 78, 9 mmHg; p=0,000); Total c (170, 6 vs. before CSII (7.2±0.9% vs. 7.6±1.0%, p<0.001). 154, 3 mg/dl; p=0, 00); LDLc (96, 7 vs. 86, 9 mg/dl; p=0, 00); HDLc (42, 3 vs. Conclusion: The use of CSII in our population in southwestern Spain is 43, 74 mg/dl; p=0, 07); TG (178, 3 vs. 140, 5 mg/dl; p=0, 05); no.OADs (2, 08 higher than the European average, achieving a signifi cant improvement vs. 0, 88; p=0,000) and insulin dose (0, 6 vs. 0, 5 U/kg/d); A1c>8% (53, 8% vs. in glycaemic control, both in children and adults. National registries are 17, 4%; p=0.000); A1c<7% (11, 0% vs. 44, 9%; p=0.000). No renal impairment needed to identify the number of patients treated with CSII and the benefi ts observed. In multivariate analysis, A1c reduction is associated to basal A1c obtained. but it is independent of other variables. Weight loss is also independent of Table 1. Demographic Characteristics. age, sex, duration, A1c, weight, OADs or BP. Conclusion: LIRA has a long-term effectiveness in A1c and weight reduc- adults children tion for the majority of patients, regardless of diabetes duration. The per- n 621 158 centage of discontinuations was low and renal function did not worsened. Gender (male%) 51 53 Age (years) 40 ± 14 12 ± 4 2254-PUB Duration of diabetes (years) 20 ± 14 5 ± 4 Cost-Effectiveness of Twice Daily vs. Insulin Glargine as Add-on Therapy to Oral Antidiabetic Agents in Type 2 Diabetes in Patients on CSII n (%) 122 (20) 42 (27) China Duration of CSII (years) 3.1 ± 3.6 (0-12) 2.8 ± 1.9 (0-6) XIAOYONG WANG, SHUYAN GU, QING QIAO, WEIGUO GAO, SUSAN GRANDY, Data are expressed as mean ± standard deviation, unless specifi ed. HENGJIN DONG, Shandong, China, Hangzhou, China, Gothenburg, Sweden, Shang- hai, China, Gaithersburg, MD Increasing prevalence and long term complications associated with poor 2252-PUB glycemic control in patients with type 2 diabetes (T2DM) impose heavy The Differences in Education on Pump Therapy between Younger burden on patients and healthcare system in China. In addition to its hypo- and Older Adults with Type 1 Diabetes glycemic effect, exenatide twice daily (EBID) can reduce weight and risk of ELEKTRA SZYMANSKA-GARBACZ, JERZY LOBA, LESZEK CZUPRYNIAK, Lodz, hypoglycemia compared with insulin glargine (IG), which are observed to Poland, Warsaw, Poland independently increase risk of cardiovascular diseases and treatment costs. However, long term treatment benefi t and cost-effectiveness of EBID in real

Therapeutics Pump therapy requires continuous patients’ education. Its objective is to world setting in China is still unknown. This study aims to estimate cost-

Clinical Diabetes/ avoid errors with pump management. As education should be tailored to

PUBLISHED ONLY each patient’s needs and capability, we conducted a study analysing whether effectiveness of EBID vs. IG in a real world setting. Cardiff Diabetes Model any age-specifi c educational needs in younger and older adults exist. The was used to simulate disease progression and estimate long term effects of study group comprised 60 type 1 diabetes patients using Medtronic 722, EBID vs. IG. Systematic literature reviews (English and Chinese databases) 754 and 640G pumps, divided in two subgroups: Group 1 - aged <26 years obtained 3 randomized clinical trials compared EBID vs. IG as add-on ther- and Group 2 - >26 years (n=29 and 31; mean age 22.6±2.3 and 34.5±8 yrs; dia- apy for T2DM in China; meta-analysis was used to combine model required betes duration 7.5±5.6 and 11.2±9.8 yrs; pump use 4.2±3.8 and 3.0±2.6 yrs, clinical data and patient profi les. Costs for T2DM patients (≥35 years) with HbA1c 7.5±5.6% and 7.2±0.6%, respectively; all p>0.05). Inappropriate car- and without complications incurred between 1/1/2012 and 12/31/2014 were bohydrate counting, dominant manual boluses use, inadequate bolus wizard extracted from a claims database in a provincial hospital in Shandong, China. overriding, delayed infusion set change or improper fi lling, unjustifi ed pump From payers’ perspective, costs (2014 U.S. $) and benefi ts were estimated disconnection episodes, boluses given without self monitoring of blood glu- over 40 years at a discount rate of 3%. Patients on EBID had lower predicted cose (SMBG), miscorrection of hypo- or hyperglycemia events, no boluses incidences of cardiovascular and hypoglycemia events and lower total before eating, no SMBG at bedtime, too rare or too frequent use of SMBG costs compared with those on IG. There were increased numbers of qual- or boluses, and improper use of temporary basal rate were noted retrospec- ity adjusted life years (QALY; 1.88) and life years (LY; 0.01) at a cost saving tively from all two weeks’ records from a year preceding the moment of data of $17709 gained with EBID compared with IG (i.e., cost saving of $9439/ collection. Mean number of errors was similar in both groups, however older QALY and $1842719/LY) per patient. Sensitivity analyses confi rmed that the patients overcorrected hyperglycemia signifi cantly more often (p<0.05) than results were robust. In Chinese T2DM patients inadequately controlled by the younger subjects, however those younger ones used SMBG (p<0.01) and oral antidiabetic agents, EBID is a cost-effective add-on therapy alternative boluses (p<0.05) less often than the older patients. Pump therapy education compared with IG, which may address an excess of medical needs attributed addressed to younger adults should concentrate on improving SMBG use to weight gain and hypoglycemia in T2DM treatment. and administering appropriate boluses, while the older patients should be Supported By: AstraZeneca encouraged to be more complacent when managing hyperglycemia. Supported By: Medical University of Lodz 2255-PUB Effectiveness and Safety of Switching from Dipeptidyl Peptidase-4 Inhibitors to GLP-1 Receptor Agonists in Patients with Type 2 Dia- CLINICAL THERAPEUTICS/NEW TECHNOLOGY— betes NON-INSULIN INJECTABLES JUAN J. GORGOJO-MARTINEZ, ALBERTO SANZ-VELASCO, GARA FEO-ORTEGA, CLARA SERRANO-MORENO, Madrid, Spain 2253-PUB The aim of this 12-month retrospective study was to evaluate the effec- tiveness and safety of switching from a dipeptidyl peptidase-4 (DPP-4) inhib- Effectiveness, Safety, and Predictors of Clinical Response to Lira- itor to a GLP-1 receptor agonist (GLP-1 RA) in patients with type 2 diabetes glutide in Type 2 Diabetes Patients with Poor Glycaemic Control: and HbA1c >6.5%. The main outcomes were changes in HbA1c and weight at 2-Year-Follow-up in Real-World Setting 1 year, rate of withdrawals (WD) and adverse effects. A logistic regression EDURNE LECUMBERRI, MAITE ORTEGA, ELENA LÓPEZ-MEZQUITA, DOMINGO analysis was performed to fi nd the best predictive model of WD. OROZCO-BELTRAN, CLOTILDE VÁZQUEZ, Madrid, Spain, Alicante, Spain 111 patients (48.6% female) were included, mean age 59.5 y, weight Objectives: To analyze the effectiveness of Liraglutide (LIRA) on fasting 99.8 kg (BMI 37.5 kg/m2) and HbA1c 8.2%; 61.3% of patients switched to plasma glucose (FPG), A1c (A1c), blood pressure (BP) liraglutide (LIRA), 19.8% to exenatide twice daily (EXE BID) and 18.9% to and profi le after a 2-year follow-up. To describe predictors of response. exenatide LAR (EXE LAR). HbA1c, weight and systolic blood pressure (SBP) To describe renal and gastrointestinal tolerance. signifi cantly decreased and glomerular fi ltration rate (eGFR) increased 1 year Methods: Observational prospective study. 24-month follow-up. Inclu- after switching to a GLP-1 RA (Table). 30.6% of patients experienced nausea, sion criteria: Patients with type 2 DM, obesity and poor glycaemic control. 17.1% vomiting and 20.7% mild hypoglycemia (patients with SU or insulin); Paired sample T-test, Chi2 and multiple logistic regression analysis were per- 27% of patients withdrew, mainly due to nausea. There were more WD due formed; statistical signifi cance p<0, 05. to all causes and nausea with EXE BID vs. LIRA or EXE LAR. In a logistic Results: 133 patients (55% male), mean age 57, 6± 10, 6 and mean dura- regression model, EXE BID, lower weight and diabetic neuropathy were inde- tion of disease 7, 9±6, 3 years. 73, 9% achieve the composite outcome of pendent predictors of WD. reducing A1c and weight; 14, 3% of treatment discontinuation (73, 3% due In summary, patients who switched to GLP-1 RA from a DPP-4 inhibitor to gastrointestinal side-effects). Basal and fi nal parameters: A1c (8, 22 vs. improved glycemic control, weight, SBP and renal function, although a sub- 7, 18%; p=0,000); FPG (170, 6 vs. 122, 2 mg/dl; p=0,000); weight (100, 9 vs. group experienced gastrointestinal intolerance, especially with EXE BID. 96, 3 kg; p=0,000); BMI (37, 2 vs. 35, 4 kg/m2; p=0,000); sBP (133, 4 vs. 127,

For author disclosure information, see page A696. ADA-Supported Research

A570 CLINICAL THERAPEUTICS/NEW TECHNOLOGY—NON-INSULINCATEGORY INJECTABLES

Table. Evolution of Patients (Completers) after Switching to a GLP-1 RA. better quality of life according to the SF-12 score (p=0.059) was observed in baseline 12 months paired difference signifi cance the fi rst group. The group of glargine and achieved a signifi cant (mean ± SEM) (mean ± SEM) (95% CI) of change (p) improvement of self-esteem according to the Rosenberg scale (p=0.033) as weight (kg) 100.7 ± 1.9 95.5 ± 1.9 5.2 (4.0-6.5) <0.0001 opposed to the rest of the groups. The positive effect of the combination of glargine and lixisenatide was achieved as early as right after the fi rst trimes- 2 BMI (kg/m ) 37.5 ± 0.76 35.5 ± 0.75 2.0 (1.5-2.4) <0.0001 ter and it was even improved right at the end of six months. fasting plasma glucose (mg/dl) 172.3 ± 6.2 155.4 ± 4.7 17.0 (3.1-30.9) 0.018 Patients under the combination of glargine and lixisenatide achieve ade- HbA1c (%) 8.31 ± 0.17 7.28 ± 0.14 1.03 (0.64-1.42) <0.0001 quate glycemic control and a signifi cant improvement in quality of life and systolic blood pressure (mmHg) 138.9 ± 2.2 134.6 ± 1.8 4.3 (0.4-8.1) 0.029 self-esteem as early as three months into the therapeutic regimen. glomerular fi ltration rate (ml/min) 75.9 ± 2.7 82.0 ± 2.6 6.2 (3.1-9.2) <0.0001 2258-PUB SEM: standard error of mean. GLP-1 Analogue Liraglutide and Detemir Insulin in NODAT Diabetes: A Four-Year Follow-Up 2256-PUB MOHAMMED EL MOKHTARI, MIRO POPESCU, Pointe-à-Pitre, Guadeloupe Real-World Outcomes Associated with for the Treat- Introduction: New-onset diabetes mellitus after transplantation (NODAT) ment of Type 2 Diabetes in a Commercially Insured U.S. Health Plan is a serious and common complication following solid organ transplantation. SARAH W. THAYER, MIRKO SIKIRICA, CHRISTOPHER BELL, MICHAEL BOGART, NODAT has been reported to occur in 4% to 25% of renal transplant recipi- AMIT BHATTACHARYYA, ASHISH V. JOSHI, Eden Prairie, MN, Collegeville, PA, ents, 2.5% to 25% of liver transplant recipients, 4% to 40% of heart trans- Research Triangle Park, NC plant recipients, and 30% to 35% of lung transplant recipients. The diagno- Real-world effectiveness analyses provide valuable data that may not sis of NODAT should be based on the 2003 updated ADA criteria for type 2 be captured in clinical trials. This retrospective claims study describes diabetes. patient characteristics and assesses glycated hemoglobin (A1c) outcomes in Objective: The objective is to assess the effectiveness of liraglutide in patients with type 2 diabetes (T2D) initiating albiglutide (ALBI), a glucagon- combination with insulin analogue detemir in reducing HbA1c, and to evalu- like 1 receptor agonist. ate safety and tolerability in NODAT. Commercial enrollees in a large U.S. health plan with ≥1 prescription Between 2006 and 2011 we had 260 renal transplant recipients in our (Rx) claim for ALBI (7/29/14-3/31/15) were included; the date of fi rst fi ll was hospital and among them 28 developed diabetes. Seventeen had normal Therapeutics the index date. Plan members were observed for 6 months prior to (base- renal function after transplantation. We have treated them with liraglutide Clinical Diabetes/

line) and 6 months after the index date (follow-up). Inclusion criteria were in combination with . PUBLISHED ONLY age ≥18, continuous plan enrollment ≥12 months, and T2D diagnosis or oral Results: The mean HbA1c after three years of treatment was 6.9% (initial antidiabetic (OAD) Rx in baseline or follow-up. Members with type 1 diabe- mean HbA1c 10.6%) with no or little hypoglycemia, no change in renal func- tes diagnosis and ≥1 insulin Rx, gestational diabetes, or pregnancy were tion and no weight gain. Two patients showed after two years of treatment excluded. Patient characteristics and Rx treatment patterns were analyzed an increased (over three fold increase) but without signs of pancreati- descriptively. A1c was assessed after ≥90 days of follow-up in the subset of tis. In these cases the treatment was stopped. patients with baseline and follow-up lab data. Conclusion: The use of liraglutide in combination with insulin detemir A total of 745 patients were identifi ed with the following baseline charac- seems to be an optimal therapeutic option with good glycemic control, fewer teristics: age 52.7±9.0 years, 54.9% male, Quan-Charlson score of 0.57±0.97, side effects and no weight gain in patients with NODAT. A regular survey of and diabetes complications severity index of 0.70±1.13. Prior to ALBI Rx, the lipase is needed. Further studies are needed with a higher number of 41.9% of patients had ≥1 OAD with insulin or other injectables; 28.1% had ≥2 patients to confi rm that. OADs without insulin or other injectables; and 13.4% had OAD monotherapy. Patients averaged 4.4±2.1 ALBI Rxs in the follow-up period. Average adher- 2259-PUB ence via proportion of days covered was 0.68. A subset of members with Comparison of Usability, Accuracy, Preference, and Satisfaction available A1c results (n=126) had 0.89% lower A1c in the follow-up period between Three Once-Weekly GLP-1 Receptor Agonist Pen Devices (8.64% vs. 7.75%; P<0.001). JENNIFER M. TRUJILLO, ANNA Y. ZHOU, Aurora, CO This fi rst ever real-world assessment of ALBI in a U.S. health plan shows Three glucagon-like peptide-1 receptor agonists (GLP-1 RAs) are avail- that members using ALBI experienced statistically signifi cant reductions in able as once-weekly injections using prefi lled, single-use, pen devices; each A1c levels 3-6 months after initiating ALBI. Further research is warranted to with different preparation and administration requirements. This was an understand long-term and comparative outcomes associated with ALBI vs. open-label, task and interview based study comparing the usability, accu- other T2D medications. racy, preference, and satisfaction between albiglutide, exenatide XR, and in 30 pharmacy student participants. Participants watched the 2257-PUB products’ instructional videos in a randomized order and then demonstrated The Effect of the Combination of Basal Insulin and a Short-Acting how to use each device to the same evaluator. Accuracy and usability were GLP-1 Agonist on the Glycemic Control, the Compliance, and the assessed using key performance measures (time taken to complete tasks, Quality of Life of Type 2 Diabetes Mellitus Patients number of user errors, completing steps in order). A participant survey ANASTASIOS G. KOUTSOVASILIS, ALEXIOS SOTIROPOULOS, MARIA PAPPA, assessed overall preferences for each product and satisfaction using a CHARALAMPOS TAMVAKOS, EVDOXIA BLETSA, ANNA GROZOU, DESPINA 5-point Likert scale (5=very easy; 1=very diffi cult). Data were analyzed using PAPADAKI, ILIAS TAMVAKOS, STAVROS BOUSBOULAS, Nikaia, Greece the Mann-Whitney-U and Kruskal-Wallis tests. Tasks were completed faster The combination of basal insulin and a GLP-1 agonist has several positive (p<0.001), with less user errors (p<0.01) with the dulaglutide pen compared effects. The aim of this study is to examine the effect of the combination of to the exenatide XR and albiglutide pens. The exenatide XR pen was better glargine and lixisenatide on the regulation, the compliance and the quality than the albiglutide pen for these parameters (p<0.001). Twenty-eight par- of life of type 2 diabetes mellitus patients compared to those of patients ticipants (93%) preferred dulaglutide in all categories including overall pref- under a regimen of basal insulin and rapid acting insulin or basal insulin and erence, ease of use, ease of demonstration, and confi dence in full dose deliv- DPP-4 inhibitor. ery (p<0.001 for all). Albiglutide was least preferred. All user satisfaction The fi rst group of patients received insulin glargine and lixisenatide (n=48), questions were higher for dulaglutide compared to albiglutide and exenatide the second group was under glargine and a rapid acting insulin (n=49), while XR (p<0.001). There were no signifi cant satisfaction differences between the third group received glargine and DPP-4i (n=52). All three groups were albiglutide and exenatide XR except for recalling all steps where exenatide also receiving . After three and six months, glycemic control XR performed better (p<0.05). Dulaglutide was associated with higher user (HbA1c), patients’ compliance to therapeutic treatment (using the Morisky satisfaction, faster demonstration with fewer errors compared to exenatide 8-Item Medication Adherence Questionnaire), quality of life (using the SF-12) XR and albiglutide. Most participants preferred the dulaglutide pen overall. and self-esteem (using the Rosenberg self-esteem scale) were assessed. Six months later, 74.4% of the patients on glargine and lixisenatide achieved the individualized target of HbA1c vs. 59.3% of patients on second group and 68.7% of the patients on third group (p=0.016). Patients on glargine and lixisenatide showed a decrease of body weight as opposed to other patient groups (p=0.033). A higher degree of compliance (p=0.026) and a

ADA-Supported Research For author disclosure information, see page A696.

A571 CLINICAL THERAPEUTICS/NEW TECHNOLOGY—NON-INSULINCATEGORY INJECTABLES

2260-PUB Table. Dulaglutide (Dula), the Once-Weekly Glucagon-like Peptide Respondes (N=202) Non responders (N=74) Receptor-antagonist (GLP-1RA), Exerts a Superior Glucose-stabiliz- Baseline 3 months Study Baseline 3 months Study t-test for the ing Effect as Compared with Liraglutide (Lira) in Hemodialysis (HD) after end after end between group Patients with Type 2 Diabetes (T2D): An Assessment by Continuous difference at each Glucose Monitoring (CGM) chech-up point; SATOSHI FUNAKOSHI, JUNICHIRO HASHIGUCHI, MARIE MAEDA, YUKIKO IWAI, p value KENJI SAWASE, HIROSHI ICHINOSE, OSAMU SASAKI, TAKASHI HARADA, YOKO A1c (%) 8.89±1.06 7.47±1.14 7.06±1.53 8.49±1.30 8.49±1.41 7.79±1.69 0.009; 0.001; 0.007 OBATA, TOMOYA NISHINO, YUTAKA MORI, KAZUNORI UTSUNOMIYA, Nagasaki, LDL- (mmol/L) 2.56±0.92 2.25±0.78 2.54±0.95 2.83±1.01 2.49±0.84 2.96±1.74 0.036; 0.027; 0.012 Japan, Komae, Japan, Tokyo, Japan HD induces plasma glucose (PG) variation, and it can cause cardiovascular HDL cholesterol (mmol/L) 1.22±0.28 1.24±0.27 1.35±0.39 1.21±0.27 1.14±0.23 1.31±0.30 0.791; 0.005; 0.475 events. In this meeting last year we had reported the superior glucose-sta- (mmol/L) 2.96±1.74 2.61±2.33 2.13±1.24 2.83±1.65 2.69±1.55 2.28±1.34 0.577; 0.784; 0.385 bilizing effects of Lira + insulin degludec (IDeg) in comparison to basal-bolus Systolic blood pressure (mmHg) 141.12±19.43 139.0±16.0 129.5±16.1 149.91±20.0 139.28±14.54 140.0±13.71 0.001; 0.895; <0.001 insulin. In this study we compared the once-daily Lira to once-weekly Dula, Diastolic blood pressure (mmHg) 85.0±9.1 82.54±7.19 80.0±9.10 90.08±11.23 84.89±8.0 85.0±5.0 0.001; 0.021; 0.001 a new, long acting GLP-1RA, both combined with daily IDeg in HD patients with T2D. Eight adults with T2D on HD who had been treated with daily IDeg (5-74 units) + Lira (0.6-0.9 mg) (IDeg+Lira) were converted to once-weekly injections of the same dose of IDeg + 0.75mg of Dula (IDeg+Dula). Dula was 2262-PUB injected on a Wednesday or a Thursday as shown in the day 1 in Figure. PG Treatment Attributes Infl uencing Preferences for GLP-1 Receptor control was monitored by CGM for 5 days, and the mean amplitude of glyce- Agonists among Injection-Naïve Patients with Type 2 Diabetes Mel- mic excursions (MAGE) was calculated before and after the conversion. litus (T2DM) The signifi cant difference in MAGE between on-HD days and off-HD days LEI QIN, STEPHANIE CHEN, EMUELLA FLOOD, BEVERLY ROMERO, MARIE DE LA in IDeg+Lira was observed, whereas this difference was cancelled out 2-4 CRUZ, ALKA SHAUNIK, SUSAN GRANDY, Gaithersburg, MD weeks after the conversion to IDeg+Dula. Four out of 8 patients reported GLP-1 receptor agonists (RA) vary in effi cacy, side effects, regimen and gastrointestinal adverse events, but were transient. Dula can exert a poten- device. The study objective was to assess attributes affecting preferences tially superior glucose-stabilizing effect as compared to Lira in HD patients for GLP-1RAs among injection-naïve patients with T2DM in the UK. Patients Therapeutics with T2D. recruited through a patient panel completed a web-based, discrete choice Clinical Diabetes/ experiment survey asking for choices between hypothetical, blinded treat-

PUBLISHED ONLY Figure. Effects of Lira and Dula Both Combined with IDeg on Glucose Varia- ment profi les refl ecting exenatide QW vs. liraglutide QD. Eight attributes were tion Assessed by CGM in HD Patients with T2D (N=8). included: effi cacy, common side effects, device size, needle size, titration, prep- aration, evidence of long-term effi cacy/safety, and dosing frequency. Analysis involved a conditional logit model, with each attribute and level included sepa- rately. Odds ratios (OR) and 95% CIs were calculated to indicate likelihood of choosing a treatment with a given attribute level vs. a reference attribute level. 297 patients completed the survey (mean age 62 y, T2DM duration 8.3 y). The 3 most infl uential attributes were effi cacy (OR for -1.5% A1c vs. -0.8%=1.75, 95% CI=1.57-1.96; p<.001); common side effects (OR for level refl ecting exenatide QW vs. liraglutide QD=3.54, 95% CI=3.28-3.82; p<.001); and dosing frequency (OR for weekly vs. daily=1.69, 95% CI=1.57-1.82; p<.001). Among device and reg- imen-related features, dosing frequency, preparation, and titration were most important. Patients prefer weekly to daily dosing, preparation for a multi-use pen to an autoinjector, a single-use pen, and vial/syringe, and no titration to titration. Needle and device size refl ecting current GLP-1RAs were not signifi - cant predictors of choice. Based on the survey, injection-naive patients favored a hypothetical GLP-1RA refl ecting exenatide QW attributes vs. liraglutide QD (OR = 6.15; p<.001) attributes, with the OR increasing to 10.8 for a hypothetical exenatide QW autoinjector vs. liraglutide QD. Patient preferences should guide shared decision-making for treatment decisions to improve adherence.

2261-PUB 2263-PUB Factors Predicting GLP-1 Receptor Agonists Effi cacy in Type 2 Dia- Impact of Baseline BMI on Glycemic Control and Weight Change betes Mellitus Treatment: A Retrospective Observational Study with Exenatide Monotherapy in Chinese Type 2 Diabetes Patients LEA DUVNJAK, KRISTINA BLASLOV, Zagreb, Croatia HONGRONG DENG, SHUO LIN, JING LV, WEN XU, SIHUI LUO, BIN YAO, JIANPING As glucagon-like peptide-1 receptor agonists (GLP-1 RA) are widely used WENG, Guangzhou, China to achieve glycaemic control in patients with type 2 diabetes, early identi- The weight reduction effect of exenatide is well proved, which may con- fi cation of GLP-1 RA responders is of special clinical interest. We evaluated cerns physicians about too much weight loss in patients with normal weight. biochemical and clinical characteristics in GLP-1 RA responders and non- This study evaluated the effects of exenatide monotherapy on glycemic responders (defi ned by switching to insulin therapy at any time after inter- control and weight change among normal-weight, overweight, and obese vention) during 28 months treatment. Among 276 subjects, 73.18% were patients with newly diagnosed T2DM. responders. Compared to non-responders they were younger (57.92 vs. 59.69 Twenty-nine normal-weight (BMI 18.5-23.9 kg/m2), 53 overweight (BMI years, p=0.001), had higher BMI (39.08 vs. 38.62 kg/m2, p=0.001), waist cir- 24.0-27.9 kg/m2), and 26 obese (BMI ≥28 kg/m2) drug-naive T2DM patients cumference (123.42 vs. 120.25 cm, p=0.001), A1c (Table), C-peptide (0.92 vs. were treated with exenatide for 48 weeks. 0.79 ng/mL, p<0.001) while shorter disease duration (10.54 vs. 13.26 years, Patients’ characteristics at baseline and after 48-week treatment were p=0.009). Improvement in cardiovascular (CV) risk factors was observed in shown in Table 1. After 48-week treatment, mean HbA1c changes were both groups, but patients with severe hypertension, dyslipidaemia and his- -2.27%, -1.72% and -1.41% in normal-weight, overweight and obese patients tory of coronary artery disease (CAD) (25.67 vs. 5.94%, p=0.001) failed to (P=0.724 among groups after adjusted for baseline values). Fasting, 0.5 h maintain glycaemic control with GLP-1 RA. An A1c reduction of 15% after 3 ≥ and 2 h postprandial plasma glucose decreased similarly among groups. The months showed the best predictive value in identifying responders to GLP-1 weight reductions presented a non-signifi cantly increasing trend from nor- RA (HR 1.55; 1.47-3.01), while the history of CAD showed negative predictive mal-weight, overweight to obese patients (decreased by 2.20 kg, 3.91 kg and value (HR 0.27; 0.185-0.403). Patients with higher CV risk factors failing to 4.01 kg, respectively, P=0.104), as well as the waist changes (decreased by reach 15% A1c decrease 3 months after the introduction of GLP-1 RA might ≥ 2.2 cm, 3.2 cm, and 5.6 cm, respectively, P=0.078). be unlikely to maintain long-term glycaemic control. Baseline BMI had no impact on glycemic control, weight change, or other metabolic index with exenatide monotherapy. Normal-weight T2DM patients would benefi ts from exenatide as much as the overweight or obese patients on glucose control with less weight reduction.

For author disclosure information, see page A696. ADA-Supported Research

A572 CLINICAL THERAPEUTICS/NEWCATEGORY TECHNOLOGY—ORAL AGENTS

Table 1. Comparison of Glucose Control, Weight Change, and Other Index at 2264-PUB Baseline and After Treatment. Effectiveness and Tolerability Profi le of GLP-1 Receptor Agonists in Variable Normal- Overweight Obese P value among groups Elderly Patients with Type 2 Diabetes weight JUAN J. GORGOJO-MARTINEZ, GARA FEO-ORTEGA, CLARA SERRANO- Gender, M/F 17/12 36/17 19/7 p=0.473 MORENO, ALBERTO SANZ-VELASCO, Madrid, Spain The aim of this 12-month retrospective single-center study was to evalu- Age (years) 52.0±11.2 50.3±8.9 45.0±9.6 normal-weight vs. overweight: p=1.000 ate the effectiveness and safety of 3 GLP-1 receptor agonists (GLP-1 RA), normal-weight vs. obese: p=0.028 overweight vs. obese: p=0.075 exenatide twice daily (EXE BID), exenatide once weekly (EXE OW) and lira- glutide (LIRA), in elderly (≥65 years) vs. younger (<65 years) patients with HbA1c (%) type 2 diabetes (T2DM). The main outcomes were changes in HbA1c and Baseline 8.4±1.2 7.9±1.0 7.9±1.0 normal-weight vs. overweight: p=0.009 weight at 12 months, rate of withdrawals (WD) and adverse effects (AE). A normal-weight vs. obese: p=0.015 multiple lineal regression analysis was used to compare HbA1c and weight overweight vs. obese: p=0.843 changes between both groups controlling for baseline confounders. After therapy# 6.2±0.8 6.1±1.0 6.4±1.0 p=0.724 112 patients ≥65 years old and 317 patients <65 years old were included Weight (kg) (Table). At one year, HbA1c, fasting plasma glucose (FPG), weight and LDL Baseline 59.5±7.1 72.7±8.2 83.2±8.9 p<0.001 (between any two groups) cholesterol (LDL-C) signifi cantly decreased and estimated glomerular fi ltra- tion rate (eGFR) increased in both groups (Table). No signifi cant adjusted After therapy# 57.3±8.1 68.8±9.0 79.2±11.2 p=0.104 differences in HbA1c, FPG, weight, LDL-C and eGFR changes were seen BMI (kg/m2) between the age groups (Table). There were no signifi cant differences in Baseline 22.2±1.3 26.0±1.1 29.9±1.2 p<0.001 (between any two groups) AE or WD between elderly and younger patients: nausea (29.5% vs. 37.9%), After therapy# 21.4±1.4 24.5±1.7 28.4±2.6 p=0.157 vomiting (16.1% vs. 19.9%), minor hypoglycemia (23.2% vs. 16.4) and WD (30.4% vs. 27.1%). There were more WD due to all causes and vomiting with Waist (cm) EXE BID vs. LIRA or EXE LAR in both younger and elderly patients. Baseline 81.8±5.9 90.8±7.4 100.7±6.0 p<0.001 (between any two groups) In summary, treatment with GLP-1 RA showed similar effectiveness and After therapy 79.6±6.5 87.7±7.5 95.1±7.7 p=0.078 tolerability in elderly and younger patients with T2DM.

p value 0.089 <0.001 <0.001 Table. Baseline Characteristics and Evolution of Elderly vs. Younger Patients Therapeutics Treated with GLP-1 RA. FPG (mmol/L) Clinical Diabetes/ Baseline 9.3±2.2 8.6±1.6 8.7±2.1 p=0.315 Patients aged Patients aged Difference PUBLISHED ONLY <65 years old ≥65 years old between After therapy# 6.8±1.2 7.0±1.8 7.2±2.1 p=0.159 n 317 n 112 groups (p) 0.5h PPG (mmol/L) age (years) 52.7 (8.8) 70.5 (4.2) <0.0001 Baseline 12.0±2.5 11.8±2.5 11.4±2.6 p=0.193 male/female (%) 46.1/53.9 41.1/58.9 0.36 duration of DM (years) * 7.0 (2.5-12.0) 16.1 (9.1-22.4) <0.0001 After therapy# 9.4±2.2 9.7±2.5 9.4±2.6 p=0.679 weight (kg) 109.5 (21.6) 91.4 (14.8) <0.0001 2h PPG (mmol/L) BMI (kg/m2) 40.7 (7.2) 36.6 (6.4) <0.0001 HbA1c (%) 7.5 (1.6) 7.8 (1.5) 0.09 Baseline 14.6±3.5 13.7±2.8 13.2±3.1 p=0.192 eGFR (ml/min) 81.9 (20.6) 62.2 (18.7) <0.0001 After therapy# 11.2±3.1 10.5±3.4 10.8±3.7 p=0.533 exenatide BID (%) 29.7 19.6 0.04 TC (mmol/L) exenatide OW (%) 13.9 22.3 0.037 liraglutide (%) 56.5 58 0.773 Baseline 4.9±1.4 5.1±1.0 4.9±1.0 p=0.621 HbA1c (%) reduction 0.81 (0.62-1.00) 0.57 (0.22-0.93) 0.588 After therapy 4.8±1.1 4.9±1.1 4.6±1.0 p=0.569 FPG (mg/dl) reduction 23.0 (14.3-31.7) 26.9 (12.7-41.1) 0.512 p value 0.729 0.063 0.036 weight (kg) reduction 5.3 (4.1-6.4) 4.1 (3.0-5.2) 0.699 TG (mmol/L) BMI (kg/m2) reduction 1.98 (1.56-2.40) 1.65 (1.18-2.13) 0.91 Baseline 1.6±1.1 2.0±1.1 2.1±1.2 p=0.141 LDL-C (mg/dl) reduction 11.1 (6.4-15.9) 11.8 (5.6-18.0) 0.085 After therapy 1.2±0.6 1.7±1.3 2.2±2.2 p=0.277 eGFR (ml/min) increase 5.4 (3.4-7.3) 3.8 (0.7-7.0) 0.176 p value 0.092 0.055 0.749 Baseline data [%, mean (SD) or *median (IQR)]; evolution [paired differences HDL-C (mmol/L) baseline vs. fi nal visit (95% CI)]. Baseline 1.2±0.3 1.2±0.3 1.0±0.2 p=0.561 After therapy 1.3±0.3 1.3±0.4 1.0±0.2 p=0.252 p value 0.134 0.022 0.962 CLINICAL THERAPEUTICS/NEW TECHNOLOGY— LDL-C (mmol/L) ORAL AGENTS Baseline 3.3±0.8 3.2±0.8 3.0±0.7 p=0.855 2265-PUB After therapy 3.1±0.9 3.0±0.9 2.9±0.8 p=0.785 Comparison of and in Patients with Type 2 p value 0.156 0.217 0.016 Diabetes in China: A Multicenter, Randomized, Controlled Clinical SBP (mmHg) Trial Baseline 125±18 126±15 126±13 p=0.928 ZI LIN SUN, YANG YUAN, YANMEI LIU, CHEN GUANG WU, JUN LIANG, SHAO GANG MA, FEI HUA, PEI LIU, Nanjing, China, Yancheng, China, Zhenjiang, China, After therapy 121±11 122±14 123±13 p=0.800 Xuzhou, China, Huai’An, China, Changzhou, China p value 0.144 0.068 0.348 Objective: To evaluate the effi cacy and safety of Mitiglinide vs. Acarbose DBP (mmHg) in patients with type 2 diabetes mellitus (T2DM). Baseline 78±10 81±10 81±9 p=0.416 Methods: In this 12-week clinical trial, patients with T2DM within 5 years were randomly assigned to mitiglinide (10mg tid) or acarbose (50mg After therapy 76±8 78±9 78±8 p=0.848 tid) treatment group. The primary outcome was HbA1c reduction, second- p value 0.268 0.036 0.084 ary outcome were fasting blood glucose (FBG), postprandial blood glucose Data are mean±SD. #p<0.05 compared with baseline. HbA1c: glycosylated (PBG) reduction. hemoglobin, BMI: body mass index, FPG: fasting plasma glucose, 0.5 h PPG: Results: A total of 248 patients were enrolled and 237 patients were 0.5 h postprandial plasma glucose, 2 h PPG: 2 h postprandial plasma glucose, fi nally included (118 in the mitiglinide group and 119 in the acarbose group). TC: total cholesterol, TG: triglycerides, HDL-C: high-density lipoprotein-cho- After treatment, HbA1c, FBG, PBG were all decreased in the two groups lesterol, LDL-C: low-density lipoprotein-cholesterol, SBP: systolic blood pres- (P < 0.0001). HbA1c reduction was (0.95±1.11)% in the mitiglinide group and sure, DBP: diastolic blood pressure. (0.80±1.27)% in the acarbose group (P>0.05). At week 8, the FBG reduction Supported By: National Science Fund for Distinguished Young Scholars was (1.31±1.29) mmol/L (mitiglinide) and (0.86±1.68) mmol/L (acarbose),

ADA-Supported Research For author disclosure information, see page A696.

A573 CLINICAL THERAPEUTICS/NEWCATEGORY TECHNOLOGY—ORAL AGENTS

there was signifi cant difference between two groups (P<0.05). At week 12, 2269-PUB the differences in reduction of FBG and PBG from baseline between the two Effect of on Non-HDL-C in Diabetic Dyslipidemia groups were not statistically signifi cance (P>0.05). The incidence of adverse PANKAJ ANEJA, New Delhi, India event was 6.54% in the mitiglinide group and 14.66% in the acarbose group Non-HDL-C is a useful predictor of adverse cardiovascular outcome in with a signifi cant difference between the two groups (P<0.05). Main side patients treated with statin and is a recommended lipid target especially effects in the acarbose group were abdominal distension, diarrhea, consti- when triglycerides are greater than 200 mg/dL. Saroglitazar is a novel dual pation and abdominal pain, and main side effects in the mitiglinide group PPARα/γ agonist approved in India for the treatment of hypertriglyceri- were diarrhea and abdominal pain. demia in type 2 diabetes not controlled by statins. This is a single centre, Conclusion: HbA1c, FBG and PBG could all be decreased signifi cantly by observational study conducted to evaluate the effect of saroglitazar on non either mitiglinide or acarbose. Treatment with mitiglinide acquired greater HDL-C at 1 year follow-up in patients with diabetic dyslipidemia. Total 81 changes in HbA1c and FBG than acarbose, but there was no signifi cant dif- patients with type 2 diabetes and dyslipidemia were prescribed Saroglitazar ference between two groups. Mitiglinide group had fewer gastrointestinal 4mg once daily for 1 year to evaluate effect on non HDL-C. The mean age reaction and better safety. (NCT02143765). of these patients was 55.43 years with 71.6% male participants. At base- Supported By: Jiangsu Chia-Tai Tianqing Pharmaceutical Co., Ltd. line, all were on one or more antidiabetic medication and 81.48% were on statins (atorvastatin 5-10 mg/d or rosuvastatin 5-10 mg/d). At 1 year follow- 2266-PUB up, saroglitazar treatment showed signifi cant reduction in non HDL-C from Use of with Intensively Titrated Insulin Attenuates a 183.10±33.81 mg/dL at baseline to 129.09±20.70 mg/dL (p =0.0001). Other Time-dependent Increase in the Rate of Hypoglycemia lipid parameters like triglycerides and HDL-C also improved signifi cantly R. RAVI SHANKAR, LEI XU, EDWARD A. O’NEILL, KEITH D. KAUFMAN, SAMUEL from 294.14±83.90 mg/dL and 41.35±4.63 mg/dL at baseline to 130.67±32.7 S. ENGEL, Kenilworth, NJ mg/dL and 45.04±5.43 mg/dL respectively at 1 year follow-up. There were In a recent study, patients (PTs) with T2DM on insulin (INS) ± Metformin no reports on weight gain or any other adverse events. Saroglitazar in dia- were treated with sitagliptin (SITA) or placebo (PBO) while intensively titrat- betic dyslipidemia has been found effective in reducing not only non HDL-C ing INS glargine. Compared with PBO, SITA led to lower incidence of AEs of but also other lipid abnormalities. symptomatic hypoglycemia (Sx HYPO; mean difference [Δ] = -11.6%), and Table. Change in Lipid and Glycemic Parameters. all HYPO ( = -15.5%), lower incremental INS dose requirement ( = -4.7 Δ Δ Laboratory parameters Baseline After 12 months P-value units), and improved glycemic control (Δ A1c = -0.45%). To determine if dif- Therapeutics Mean±SD Mean±SD ferences in the time course of HYPO events occurred over the study period, Clinical Diabetes/ Non-HDL-C (mg/dL) 183.10±33.81 129.09±20.7 0.0001

PUBLISHED ONLY we conducted a post hoc analysis of event rates of HYPO AEs (number of HYPO AEs/PT-Year) and Sx HYPO (number of Sx HYPO/PT-Year) in the two Triglycerides (mg/dL) 294.14±83.90 130.67±32.7 0.0001 treatment groups in the primary study. For PBO there was a generally con- Total cholesterol (mg/dL) 224.88±29.98 174.83±29.98 0.0001 sistent increase in event rates of Sx HYPO AEs in each 3-week period from HDL-C (mg/dL) 41.35±4.63 45.04±5.43 0.0001 weeks 0 to 3 through weeks >21 to 24, while for SITA, although event rates of Sx HYPO AEs varied over time, they were generally consistent throughout HbA1c (%) 8.133±0.701 6.877±0.308 0.0001 the study (Figure 1). Changes in event rates of all HYPO generally paralleled FPG (mg/dL) 155.13±25.04 114.59±9.79 0.0001 the changes in event rates of Sx HYPO in each group. These analyses sug- PPG (mg/dL) 253.52±75.91 148.74±14.85 0.0001 gest that the patient burden of HYPO may not be fully refl ected in an analy- P-values are calculated using paired t-test. sis of overall incidence in studies where INS is intensively titrated (± other agents); event rates of HYPO over time may be a more appropriate index of this burden. 2270-PUB Figure 1. Event Rate of Symptomatic Hypoglycemia AE Over Time. WITHDRAWN

2271-PUB Fox01/Peroxisome Proliferator-activated Receptor γ (PPAR-γ) Net work Regulate Islet Beta-Cell Proliferation and Differentiation Depending on Glucose Level EUN JU LEE, YOU JEONG KIM, HYE SOOK JUNG, TAE KYOON KIM, TAE NYUN KIM, MIN JEONG KWON, SOON HEE LEE LEE, BYUNG DOO RHEE, JEONG HYUN PARK, MI-KYUNG KIM, Busan, Republic of Korea The direct effect of PPAR-γ on islet beta-cell is unclear yet. Previous stud- ies reported pancreatic beta-cell-specifi c ablation of PPAR-γ showed β-cell hyperplasia and increased β-cell mass. Recently, inhibits pan- creatic beta-cell proliferation and increased differentiation in near normal glucose level. PPAR-γ is one of downstream effectors of Fox01 protein in islet β-cells and Fox01 activity would be changed by glucose level. There- fore, we compared the effect of PPAR-γ agonist on beta-cell proliferation and differentiation depending on glucose level and evaluated the involve- ment of Fox01 protein. We cultured INS-1 cell in 8.3mM glucose (NG) and 33mM glucose (HG), with or without troglitazone, separately. Beta-cell proliferation was assessed by BrdU positive cell percentage. We checked Supported By: Merck & Co., Inc. phosphorylated Akt, Erk, Pax4, MafA, PDX1 and NKX 6.1 mRNA and pro- tein expression in NG and HG. Glucose induced insulin secretions were mea- sured. SiRNA for PPAR-γ and Fox01 were applied and evaluated changes of 2267-PUB factors related to proliferation and differentiation. Troglitazone decreased beta-cell proliferation in NG, but increased in HG. Phosphorylated Akt and WITHDRAWN Erk were increased by troglitazone in HG. Pax4, MafA, PDX1 and NKX 6.1 mRNA and protein expression were decrased in HG and restored by trogli- tazone. They were decreased by siRNA of PPAR-γ. SiRNA of Fox01 changed 2268-PUB pAkt, pErk and PDX1 differently depending on glucose level. In conclusion, PPAR-γ agonist directly increased beta-cell proliferation in hyperglycemia WITHDRAWN while decreased in normoglycemia. Beta-cell differentiations by PPAR-γ agonist were more increased in NG. This discrepancy may be related to Fox01/ PPAR-γ networking.

For author disclosure information, see page A696. ADA-Supported Research

A574 CLINICAL THERAPEUTICS/NEWCATEGORY TECHNOLOGY—ORAL AGENTS

2272-PUB 2274-PUB A1c Change in Canaglifl ozin Users in a Majority Medicare Managed Relationship between Baseline HbA1c, Change in HbA1c, and Clini- Care Population cal Outcomes in Type 2 Diabetes: A Comparison of Clinical Guide- ROBERT A. BAILEY, PHIL SCHWAB, YIHUA XU, ANDREW RENDA, MARGARET lines and Real-World Data PASQUALE, Raritan, NJ, Louisville, KY PHIL MCEWAN, JASON P. GORDON, MARC EVANS, RICARDO VIANA, PÄIVI M. Randomized controlled trials have demonstrated the effi cacy of canagli- PALDÁNIUS, Cardiff, United Kingdom, Basel, Switzerland fl ozin (CANA) [FDA approval March 2013] in type 2 diabetes (T2DM); real Clinical guidelines developed using indirect treatment estimates may pro- world studies measuring its effectiveness, especially in seniors, are limited. vide misleading perspectives on the relationship between baseline HbA1c This study observed A1c values in T2DM patients among Humana’s fully and expected therapy-related changes in HbA1c in patients with type 2 insured Medicare and Commercial members before and after initiating treat- diabetes (T2DM). Using patient-level data (PLD) from EDGE, a prospective, ment with CANA. Humana members are mostly Medicare. Patients were 1-year, worldwide, real-life study, and recently published clinical guidelines identifi ed by a prescription fi ll for CANA, 4/1/2013-12/31/2014, and T2DM for management of T2DM in the UK, we contrasted guideline vs. real-world defi ned as ≥ 1 inpatient stay or ≥ 2 outpatient visits with ICD9: 250.xx, (EDGE) derived estimates of HbA1c change at 1 year for Metformin + sul- excluding patients with a diagnosis for type 1 diabetes (ICD9: 250.x1 or 250. fonylurea (M+S) compared with Metformin + (M+V) in patients x3). Diabetes severity was assessed by the Diabetes Complication Severity inadequately controlled on Metformin. Multivariate analysis of EDGE PLD Index (DCSI). The date of fi rst claim for CANA was assigned the index date provided baseline HbA1c adjusted 1-year changes in HbA1c; these were con- and ≥12 months pre- and ≥6 months post-index continuous enrollment was trasted with the guideline developed equations for adjusted HbA1c treat- required. All patients had A1c result (s) < 120 days pre-index and A1c result (s) ment changes. The two approaches were assessed with respect to expected within 30 to 365 days post-index. For A1c measurement, pre-index A1c was lifetime incidence of T2DM-related vascular complications, expressed as the most recent prior to the index date. All post-index A1c measures were changes in patient life expectancy (LE) and quality-adjusted life expectancy averaged. The sample (n = 1,562) was 50% female with mean age [standard (QALE) using the CORE diabetes model. The EDGE-based estimated change deviation (SD)] of 61.8 (10.4) years and majority Medicare (61.7%). Mean (SD) in HbA1c at 1 year for M+V and M+S was -1.33% and -1.05%, respectively DCSI was 1.7 (1.9). Mean (SD) post-index enrollment was 348 (135) days. (delta: -0.28%). Applying the guideline equations, estimates for M+V and For the index CANA, 59% took 100mg, 40% took 300mg, 1% took a fi xed- M+S were -0.95% and -0.99%, respectively (delta: 0.04%). These alterna- dose combination of canaglifl ozin plus Metformin [FDA approval August tive predictions of change in HbA1c translated to a difference in the pre- 2014]. Mean (SD) days from pre-index A1c to index date was 37 (31). Mean dicted change in QALE of 0.10 years between arms (M+V vs. M+S: 0.41 vs. (SD) days from index date to post-index A1c was 185 (78). Mean (SD) A1c 0.31 years) based on the EDGE data, compared with 0.01 years (M+V vs. Therapeutics decreased from 8.6 (1.5) pre-index to 7.9 (1.3) post-index (P<.0001). The pro- M+S: 0.31 vs. 0.30 years) for the guidelines analysis; a similar relationship Clinical Diabetes/ PUBLISHED ONLY portion with A1c < 7% increased from 10.3% pre-index to 23.7% post-index was observed for LE. Gains in LE and QALE favoring M+V were driven by (P <.001). The proportion with A1c < 8% increased from 37.2% to 60.3% (P lower cumulative incidence of major microvascular and cardiovascular com- <.001). The proportion with A1c > 9% decreased from 35.0% to 16.5% (P plications. Analysis of real world data suggests that current UK guidelines <.0001). These results demonstrate statistically signifi cant A1c improvement for management of T2DM might underestimate the HbA1c lowering potential associated with canaglifl ozin treatment in a real world setting. and health gains of M+V compared with M+S in routine clinical practice. Supported By: Novartis 2273-PUB Comparison of / Fixed-Dose Combination 2275-PUB (APC) with DPP-4 Inhibitor Monotherapy in Reducing HbA1c Tissue-specifi c Modulation of PPARγ, the Key Basis of Selective DEBRA F. LAWRENCE, KEITH A. SZYMANSKI, ANDREW J. KLINK, JAY VISARIA, PPARγ Modulator INT131 in Mediating Insulin-Sensitizing Effect CLAIRE DYBALA, Deerfi eld, IL, Wilmington, DE XINNI XIE, WEI CHEN, HONG QIAO, NING ZHANG, LONG LONG, HUA LI, LILI APC is prescribed with diet and exercise to improve blood glucose con- WANG, Beijing, China trol (HbA1c) in adults with type 2 diabetes (T2D). Patient characteristics and Selective PPARγ modulator (sPPARγM) retains insulin sensitizing activity changes in HbA1c levels in patients initiating APC are assessed. Patients but with fewer side effects as the traditional PPARγ full agonist, is thought aged 18-64 years with T2D initiating APC (cases), or DPP-4 inhibitor mono- to be a promising new generation of safer insulin sensitizer. However, the therapy (controls) between 01/01/2013 and 06/30/2014 were identifi ed underlying pharmacological mechanisms of sPPARγM remain unclear. To from HealthCore Integrated Research Environment. Patients were continu- address this, a comparative study focused on target tissues effects and ously enrolled for 6 months pre and post initiation and 1:4 propensity score molecular regulation, with sPPARγM INT131 as the probe and the PPARγ full matched. Changes in HbA1c from pre to post initiation were calculated for agonist (,) as a reference drug, were performed in db/db and both groups. Of 120 cases matched with 480 controls, mean patient age was DIO mice respectively. 30mg/kg INT131 displayed equivalent insulin sensiti- 52 years, and 34% were female. Baseline diabetes severity and comorbidi- zation to 10mg/kg rosiglitazone in db/db and DIO mice, demonstrated by the ties were similar with more than 2/3 of the patients having hypertension and OGTT, ITT and HOMA-IR index. Under such circumstance, INT131 signifi cantly dyslipidemia. APC cases had more T2D-related emergency room (ER) visits increased brown adipose tissue (BAT) mass and enlarged BAT adipocyte size (10.0% vs. 5.0%, p=0.0391) and offi ce visits (2.2 vs. 1.9, p=0.0202) prior to like rosiglitazone, but it upregulated the expression of PPARγ target genes initiation vs. controls. Adherence to the index medication was similar across associated with energy expenditure and lipid more robustly in groups (mean medication possession ratio: 0.77 vs. 0.76). Pre initiation labo- BAT and muscle, exhibited more potency on inducing subcutaneous white ratory results were available for 33 cases and 140 controls (mean HbA1c adipose tissue (sWAT) browning, diminishing adipocyte size, and increas- was 9.1 vs. 8.0, p=0.0021). Among those, 17 cases and 97 controls had post ing BAT-specifi c genes expression. In contrast to rosiglitazone, INT131 didn’t initiation lab results available. Changes in HbA1c from pre to post initiation induce hepatomegaly and hepatic steatosis in both mice models. Pharma- for cases was 8.95% to 7.79%, while it was 8.01% to 7.77% for controls cokinetic analysis showed no difference on the plasma concentration of (- 1.2% vs. -0.2%, p=0.0105). In this study, baseline characteristics of HbA1c, INT-131 and rosiglitazone under equivalent doses, but the absolute content T2D-related ER visits, and offi ce visits differed between cases and controls, of INT-131 in target tissues specifi cally in adipose were remarkably higher. and sample size was limited due to lack of baseline HbA1c levels. Still, ini- In vitro coactivator recruitment assay indicated unlike rosiglitazone, INT131 tiation of APC was associated with a greater reduction in HbA1c levels than less recruit coactivator MED1, the vital mediator in PPARγ induced hepatic DPP-4 inhibitor monotherapy within 6 months of initiation. Further studies steatosis. In consistent with this, genes related to hepatic adipogenesis and should evaluate change in HbA1c over longer follow-up and across differ- lipogenesis was less induced by INT131. Current results revealed Tissue- ent treatments. specifi c modulation by PPARγ as the key basis of sPPARγM INT131 in mediat- ing insulin sensitizing effect, while with less side effects. Supported By: China Postdoctoral Science Foundation (2013M532213); National Science and Technology Major Project of the Ministry of Science and Technology of China (2012ZX09301003-001, 2012ZX09301003-003); National Natural Science Foundation of China (81102308)

ADA-Supported Research For author disclosure information, see page A696.

A575 CLINICAL THERAPEUTICS/NEWCATEGORY TECHNOLOGY—ORAL AGENTS

2276-PUB 2279-PUB Benefi cial Effects of SGLT2 Inhibitors Are Related to Serum Adi- WITHDRAWN ponectin Level in Patients with Type 2 Diabetes MASATAKA KUSUNOKI, YUKIE NATSUME, DAISUKE SATO, TAKAO NAKAMURA, YOSHIHARU OSHIDA, Nagoya, Japan, Yamagata, Japan SGLT2 inhibitor enhances glucose excretion into urine and improves gly- 2280-PUB cemic control. In addition, it has been reported that the drug has benefi cial Quality of Life Assessment in Patients with Type 2 Diabetes: Results effects on serum and body weight gain in patients with type 2 diabe- from a One-Year Observational Study in Real-World Russian Clini- tes. We hypothesized that these effects might be attribute to reduction of cal Settings body fat mass, and that the reduction may lead to elevation of ALEXEY ZILOV, AIDA EMIROVA, Moscow, Russian Federation secretion and improvement of insulin resistance. Type 2 diabetes mellitus (T2DM) is a progressive disease characterized In the present study, 95 Japanese patients with type 2 diabetes were by chronic hyperglycemia due to defects in insulin secretion and resistance. administered luseoglifl ozin (2.5 mg/day) or dapaglifl ozin (5 mg/day) for Various glucose-lowering agents are associated with adverse events, such 6-month, and were defi ned as decreased (D group; n=30) and increased (I as hypoglycemia or weight gain, which could impact patient’s Quality of Life group; n=65) serum adiponectin level at the end of the administration period. (QoL). Currently, in Russia, data on QoL assessments upon treatment with We then retrospectively evaluated the effects of the drugs on glucose- and dipeptidyl peptidase-4 (DPP-4) inhibitors are limited. lipid-metabolic parameters and renal and hepatic functions. We conducted a one-year, observational, real-world study in Russia to Six-month after the administration of SGLT2 inhibitors, while body weight assess the effectiveness and safety of vildagliptin as an add-on therapy and serum uric acid were lowered in the D and I groups, the drugs had no to Metformin or a fi xed combination of vildagliptin + Metformin in patients effects on estimated glomerular fi ltration rate and serum and LDL with T2DM (N=5231). Patients completed the EuroQoL-5D (EQ-5D) question- cholesterol. In the D group, the drugs signifi cantly elevated HbA1c and blood naire at baseline and at the end of the observation period. The question- glucose. On the other hand, in the I group, the drugs signifi cantly reduced naire consists of 2 parts. Part 1 has 5 dimensions to describe the individual’s HbA1c, blood glucose, HOMA-IR, visceral fat area, and serum aspartate mobility, self-care, usual activities, pain/discomfort, and certain psychologi- aminotransferase, alanine aminotransferase and γ-glutamyltransferase, cal problems. Patients rated problems on a 5-point scale, used to determine respectively, and serum HDL cholesterol was signifi cantly raised. Also, basal the EQ-5D score. Part 2 was a 100 point visual analogue scale on patient’s metabolic rate was tended to be elevated in the I group. Although blood urea state of health (VAS; 0, worst health; 100, best health). Results of EQ-5D Therapeutics nitrogen signifi cantly increased, there is no signifi cant difference between

Clinical Diabetes/ showed improvement in QoL of patients treated with vildagliptin on all the the D and I groups at the end of the administration period. PUBLISHED ONLY questionnaire dimensions. The proportion of patients reporting a score of These results suggest that the body weight loss induced by SGLT2 inhib- one (suggests no problem) increased from baseline to end of study: mobil- itors may not be relevant to serum adiponectin level, but that the eleva- ity (66.4% to 73.9%), self-care (87.5% to 91.7%), usual activities (65.3% to tion of adiponectin level after the administration of SGLT2 inhibitors may 79.0%), pain/discomfort (58.0% to 69.1%), and anxiety/depression (61.1% to be involved with visceral fat reduction, increase in energy expenditure, and 79.3%). Patients’ self-assessment of state of health on VAS also improved; amelioration of glycemic control and hepatic function. mean±SD score increased signifi cantly from baseline (69.13±15.45) to the end of observation period (80.70±12.59; Δ=11.59±12.15 [95% CI: 11.21, 11.97)]. 2277-PUB Based on this long-term observational study, we conclude that treat- WITHDRAWN ment with vildagliptin as add-on to Metformin signifi cantly improves health related QoL in patients with T2DM in Russia. Supported By: Novartis Pharma LLC

2278-PUB 2281-PUB Do Persistence Rates Vary between Dipeptidylpeptidase-4 Inhibi- Combination of Oral Hypoglycemic Agents and Hypoglycemia: Data tors? Mining of the Japanese Adverse Drug Event Report Database ANDREW P. MCGOVERN, WILLIAM HINTON, NEIL M. MUNRO, MARTIN B. TAKESHI HORII, YUSUKE KABEYA, SAEKA IIJIMA, MASUOMI TOMITA, TAKESHI WHYTE, SIMON DE LUSIGNAN, Guildford, United Kingdom KATSUKI, JYUNICHI SHIMIZU, YOICHI OIKAWA, Tokyo, Japan A range of dipeptidylpeptidase-4 (DPP-4) inhibitors are available for use It is often diffi cult to achieve an optimal glycemic target using a single in type 2 diabetes (T2DM) and have similar effi cacy and safety. Differing oral hypoglycemic agent (OHA). In such cases, combination therapy of sev- side effects and other medication factors affect persistence within a medi- eral OHAs is used for improving glycemic control with minimizing hypoglyce- cation class. Persistence with medication in T2DM is unsurprisingly asso- mia. However, it remains unclear what type of combination is at high-risk for ciated with better glycemic control and should be a factor in selection of hypoglycemia because of scarcity of published trials of combination therapy. DPP-4 inhibitor. We compare persistence across DPP-4 inhibitors in clinical The Japanese Adverse Drug Event Report database (JADER) contains infor- practice. mation on adverse event and medication error reports. We investigated the We performed a retrospective cohort analysis of people with T2DM using association between the type of combination therapy and risk of hypoglyce- a large UK primary care sentinel network (Royal College of General Practitio- mia using the JADER. ners Research and Surveillance Centre). We compare persistence between From 2004 through 2015, 586,844 events were reported in the JADER. The DPP-4 inhibitors from fi rst availability to July 2015. Discontinuation was Medical Dictionary for Regulatory Activities Preferred Terms was used to defi ned as a gap between prescriptions of ≥90 days. We calculated hazard identify hypoglycemia, by which 5750 reports were found in the JADER. We ratios (HR) for discontinuation using Cox regression, adjusting for patient calculated the reporting odds ratio (ROR), which is a measure of dispropor- demographics, clinical characteristics, and co-morbidities. tionality used for the purpose of detecting signals in spontaneous adverse From 58,717 people with T2DM we identifi ed 9,589 prescribed one or drug event reporting databases. The ROR is the ratio of the odds of reporting more DPP-4 inhibitors (347 prescribed alogliptin, 1,429 , 666 saxa- of one specifi c event vs. all other events for a given drug compared to the gliptin, 6,475 sitagliptin, and 144 vildagliptin). Alogliptin was excluded from reporting odds for all other drugs present in the database. analysis due to a short follow-up duration. Crude median persistence was The following combination therapies had signifi cantly elevated RORs; 2.00 years (95% CI 1.79-2.70) for linagliptin, 1.63 (1.26-2.01) for , SU+α-GI (50.7; 95% CI 40.0-64.3), SU+BG (30.6; 95% CI 20.8-44.9), 1.49 (1.42-1.55) for sitagliptin, and 0.90 (0.67-1.35) for vildagliptin. Adjusted Glinides+α-GI (28.0; 95% CI 15.1-51.8), SU+TZD (15.6; 95% CI 10.6-23.0), HRs for discontinuation per year compared to sitagliptin were 0.83 (0.74- Glinides+DPP-4I (11.8; 95% CI 5.4-25.8) and SU+DPP-4I (11.6; 95% CI 8.8- 0.92; p<0.001) for linagliptin, 0.94 (0.84-1.06; p=0.326) for saxagliptin, 1.40 15.3) showed high ROR. On the other hand, BG+ DPP-4I showed a non-signif- (1.14-1.71; p=0.001) for vildagliptin. Male gender, younger age, Asian and icant and modest ROR (2.3; 95% CI 1.0-5.3). black ethnicity, and higher BMI were associated with reduced persistence. This study shows that the combination therapies including SU or Glinides A recorded diagnosis of chronic kidney disease stage 3-5 was not associ- are at high-risk for hypoglycemia. Meanwhile, the combination of BG and ated with persistence. DPP-4I, which is now widely used in Japan, seems a safe option as combina- Persistence with linagliptin was signifi cantly longer than sitagliptin. Saxa- tion therapy. The result obtained in the Japanese database could be appli- gliptin had an intermediate level and vildagliptin, noting small numbers, the cable to patients with diabetes in Asia since they have common properties lowest persistence. of low capacity for insulin secretion.

For author disclosure information, see page A696. ADA-Supported Research

A576 CLINICAL THERAPEUTICS/NEWCATEGORY TECHNOLOGY—ORAL AGENTS

2282-PUB this randomized, double-blind, placebo-controlled crossover study, we exam- Allopurinol Changes Blood Pressure but No Uric Acid in Individuals ined the effect of , a DPP-4 inhibitor, on the postprandial glucose and with Metabolic Syndrome lipoprotein metabolism in 10 patients with type 2 diabetes inadequately con- THALES ISHIZAKI, LUCIANO GIACAGLIA, ROSA FUKUI, MARIA ELIZABETH R. trolled with oral antidiabetes medications or lifestyle modifi cation. Gemigliptin SILVA, ROSA F. SANTOS, São Paulo, Brazil or placebo was randomly added for 4 weeks and switched to the other treat- Hypertension is a risk factor for cardiovascular disease, and it has been ment with 2 weeks of wash-out period. At the end of each phase, the subjects shown that uric acid (Au) plays a role in the development of hypertension. underwent a high fat meal tolerance test. Gemigliptin signifi cantly decreased Allopurinol (AL) is a good option for treatment, it inhibits xan- the fasting and postprandial glucose levels and increased active GLP-1 levels. thine oxidase and reduces Au synthesis, it also inhibits renina angiotensin Gemigliptin signifi cantly decreased the fasting, peak and area under the curve (RA) and increases nitric oxid (NO). Previous study has shown that decrease of apolipoprotein B48 levels (Figure 1). The level tended to decrease. in blood pressure (BP) and Au were associated with AL treatment. The aim However, apolipoprotein B100 levels showed no signifi cant change. In conclu- was evaluate whether AL has independent effect on BP besides uric acid. sion, a 4-week treatment with gemigliptin effectively suppressed the postpran- Prospective pilot study was conducted among 14 adult men with metabolic dial intestinal (but not hepatic) lipoprotein secretion, possibly through the aug- syndrome (MS) divided into two groups according to Au lower or higher than mentation of active GLP-1 levels in type 2 diabetes patients. 7.0 mg/dl: G1 (Au ɖ7.0 mg/dl) and G2 (Au ≥ 7.1 mg/dl). Patients were treated Figure 1. Plasma Levels of A) Total Cholesterol, B) Triglyceride, C) Apolipo- with AL 300 mg/day for 8 weeks. The parameters were assessed pre and protein B48, and D) Apolipoprotein B100 during the High Fat Meal Tolerance post treatment. Statistical Analysis Two Way ANOVA, P <0.05. Results were Test. before and after treatment per group: Group 1 - age (54 ± 4) ys; Au (6.07 ± 0.37 X 5.35 ± 0.83) mg/dl - NS; body mass index - BMI (33.14 ± 1, 6 X 33.23 ± 1, 7) NS; waist circumference - WC (108.5 ± 1.0 x 108.8 ± 1.1) cm - NS ; systolic blood pressure (SBP) mmHg (142.0 ± 10.6 x 131.7 ± 5.6) NS; diastolic Blood pressure (DBP) mmHg (90.1 ± 2.7 x 84.1 ±1.9) P <0.03; fasting blood glucose (FBG) (100.5 ± 4.2 x 101.0 ± 3.5) mg/dl- NS; fasting blood insulin (FBI) (17.6 ± 1.4 X 19.6 ± 2.9) µU/dl - NS; A1c% (5.5 ± 0.2 vs. 5.8 ± 0.2) - NS; HOMA IR (4.39 ± 0.44 x 4.86 ± 0.67) NS and Group 2: age (50 ± 6) ys; Au (8.9 ± 0.8 x 6.9 ± 0.6) mg/dl, P <0.06; BMI (31.2 ± 2.6 x 31.0 ± 2.6) NS; WC (110.4 ± 6.6 X 109. 6 ± 6.8) cm - NS; SBP (139.0 ± 4.1 x 134.2 ± 3.9) mmHg - NS, DBP (91.2 ± 4.4 Therapeutics x 86.4 ± 1.9) mmHg, NS; FBG (97.6 ± 2.6 x 100.0 ± 3.6) mg/dl - NS; FBI (19.0 ± Clinical Diabetes/ PUBLISHED ONLY 3.2 x 17.7 ± 4.4) µU/dl - NS; A1c% (5.6 ± 0.1 x 5.5 ± 0.1); HOMA IR (4.69±0.8 x 4.38±1.1 NS). We concluded that treatment with AL: 1 - AL reduced DBP without changing Au; 2 - BP and Au reductions were not associated; 3 - The data suggest that AL do not need Au to affect BP; 4 - The AL effect on BP could be due to its action on RA and NO.

2283-PUB *P value <0.05. The Multiple Antidiabetic Actions of Musanga Cecropioides, a Plant Supported By: LG Life Sciences Used to Treat Diabetes in the Democratic Republic of Congo WEBE C. KADIMA, ALITA NICHOLS, JESSE VANUCCHI, Oswego, NY Antidiabetic properties of the crude aqueous extract of the bark of 2285-PUB Musanga cecropioides (Mc) were investigated through pre-clinical and in- Safety of Ipraglifl ozin in the Elderly: Results of a Postmarketing vitro studies. Pre-clinical testing on a small sample of people with type 2 Survey (STELLA-ELDER) diabetes demonstrated that the extract curbs high glucose excursions after YASUO TERAUCHI, KOUTARO YOKOTE, ICHIRO NAKAMURA, HARUKO SUGA- a simple meal at doses between 293 mg and 600 mg. The average per- MORI, Yokohama, Japan, Chiba, Japan, Tokyo, Japan cent increase of blood glucose for 36 subjects after ingestion of 293 mg In this postmarketing survey, the safety of ipraglifl ozin (IP) was investi- extract, 45±22%, is signifi cantly lower than in a negative control experi- gated in elderly patients (≥65 years at the start of treatment) with type 2 ment, 75±22% (p=0.0007). diabetes mellitus who were fi rst prescribed IP within 3 months of its launch Inhibition studies of involved in pathways producing glucose in- in Japan (April 2014). Of 8,687 registered patients, 7,170 patients were vivo have demonstrated that the extract of Mc inhibits alpha-glucosidase included in the safety analysis (49.2% males), 98.8% of whom were out- patients. The mean age, duration of diabetes, BMI, and estimated glom- and glucose-6-phoshatase (G6Pase). Inhibition of alpha-glucosidase was 2 investigated through kinetic studies of the hydrolysis of the substrate, para- erular fi ltration rate were 72.2 years, 10.5 years, 26.89 kg/m , and 70.2 mL/min/1.73 m2, respectively. Overall, 79.4% of patients had complica- nitrophenyl-α-D-glucopyranoside (PNPGP). Results obtained at 0.8 mM PNPGP revealed a decrease of the specifi c activity of alpha-glucisidase from tions. were normal in 86.8% of patients. Concomitant 46±4 µmole .min-1.mg-1, in the absence of inhibitor, to 25±1 µmole .min-1.mg-1, antidiabetic medications (average number 2.0/patient) were used in 83.1% and 10.9±0.3 µmole .min-1.mg-1 in the presence of the extract of Mc (10 µg/mL) of patients. There were 898 adverse drug reactions (ADRs) in 721/7,170 and Acarbose (0.80 mM), respectively. patients (10.06%). ADRs in ≥0.4% of patients included pollakiuria (0.98%), Using ammonium molybdate, the inhibition of G6Pase was monitored via pruritus genital (0.78%), rash (0.60%), thirst (0.52%), dizziness (0.52%), and the appearance of an absorption band centered at 315 nm due to the forma- dehydration (0.42%). ADRs related to genital infection, urinary tract infec- tion of a molybdate- complex. Results obtained show that 5 µg/ tion, polyuria/pollakiuria, volume depletion (e.g., dehydration), and skin com- mL of the Mc extract displays an inhibitory effect comparable to 805 µM of plications occurred in 1.45%, 0.77%, 1.32%, 1.90%, and 2.23% of patients, Metformin. respectively. ADRs related to hypoglycemia occurred in 0.32% of patients. Furthermore, Mc also exhibits anti-oxidant activity comparable to that of Of 898 reported ADRs, 443 (49.3%) occurred within <30 days of starting ascorbic acid at equal weights per volume. And, preliminary experimental treatment. The incidence rates of polyuria/pollakiuria, volume depletion, and data, suggest that phosphenolpyruvate carboxykinase (PEPCK) and glycogen skin complications within 30 days of starting treatment were 66.0% (64/97), phosphorylase are also inhibited by the extract of Mc. 49.7% (84/169), and 60.0% (102/170), respectively. Most (92.1%) events were classifi ed as recovered or improved. Skin complications did not include any serious diseases, such as Stevens-Johnson syndrome or toxic epidermal 2284-PUB necrolysis. These values are the results of an interim analysis at 6 months. In Effects of Gemigliptin, a Dipeptidyl Peptidase-4 Inhibitor, on Post- summary, the ADRs observed in elderly patients in this survey were similar prandial Lipid Metabolism in Patients with Type 2 Diabetes to those observed in general population, pre-approval clinical trials. CHANG HO AHN, EUN KY KIM, LEE KYUNG KIM, SE HEE MIN, JAE HYUN BAE, Supported By: Astellas Pharma Inc. KYONG SOO PARK, YOUNG MIN CHO, Seoul, Republic of Korea Type 2 diabetes is commonly associated with postprandial hypertri- glyceridemia (PPHTG) which is partly mediated by increased intestinal secre- tion of triglyceride-rich lipoproteins. Dipeptidyl peptidase-4 (DPP-4) inhibitors have been reported to decrease PPHTG by increasing active GLP-1 levels. In

ADA-Supported Research For author disclosure information, see page A696.

A577 CLINICAL THERAPEUTICS/NEWCATEGORY TECHNOLOGY—ORAL AGENTS

2286-PUB 2289-PUB Antidiabetic Activity of a Novel Metformin- Salt Evidence of Teneligliptin as an Add-on to the Ongoing Antidiabetic TODD LEFF, OM GOEL, Detroit, MI, Ann Arbor, MI Therapy: An Initial Real-World Experience Metformin as its hydrochloride salt is widely prescribed for the treatment RAJEEV CHAWLA, AASTHA CHAWLA, SHALINI JAGGI, VINAY KUMAR, DIPTI of type 2 diabetes mellitus. Most common adverse effect of Metformin YAKSH, NAVNEET WADHWA, New Delhi, India, Mumbai, India hydrochloride is gastrointestinal upset. 2, 4-Thiazolidinedione is an integral, We retrospectively analysed the effectiveness of teneligliptin, the recent covalently linked, structural moiety in a class of potent antidiabetes drugs - DPP inhibitor, amongst 73 continuous users, in real world clinical practice the (TZDs). The Metformin salts (1:1 or 1:2) of 2, 4-thiazoli- setting by physician diabetes specialists (I-TREAT). The baseline glyce- dinedione itself, have not been reported or studied in the treatment of type2 mic parameters, demographics, ongoing therapy and impact on the glyce- diabetes. The antidiabetic activity of a Metformin-thiazolidinedione 1:1 salt mic control were analysed through the records of patients (43 males, 30 (Met-TZ) was examined in female db/db mice. Mice received either vehicle, females), who fi lled at least one teneligliptin prescription between July- 100 mg/kg/day of Metformin, or a molar equivalent dose of Met-TZ (149 mg/ December 2015 (fi rst 6 months teneligliptin was available in India). Mann kg/day) by oral gavage. Fasting blood glucose and body weights were mea- Whitney test was used for statistical analysis. At teneligliptin initiation, sured three times per week for 4 weeks. At the end of the treatment period 31.5% used teneligliptin concomitantly with three or more other antihyperg- a glucose tolerance test (GTT) was performed. Met-TZ treatment showed lycemic agents. Metformin, sulphonylurea, pioglitazone, basal insulin, insu- striking glucose lowering activity from day 17 until the termination of the lin, SGLT 2 inhibitors continued in 93%, 89%, 34%, 11%, 8%, 4% patients, study. Met-TZ was more effective at lowering blood glucose than the equiv- respectively. 2 patients switched from other gliptins, 4 were drug naïve. At alent dose of Metformin (52% reduction vs. 43% reduction). An oral glucose 3 months follow up of 20% (15) patients, the HbA1c reductions mean ± SD tolerance test performed at the end of the study demonstrated that Met-TZ (-1.4 ± 0.63) were statistically signifi cant (baseline 9.8 ± 0.4 vs. at 3 months treatment signifi cantly improved glucose tolerance compared to vehicle, and 8.4 ± 0.4; 95% CI -2.721 to -0.1318; p=0.032). Longer follow up and future to a greater degree than Metformin alone. There was no difference in body research on larger real world longer term outcomes is warranted. Teneliglip- weight between the groups, and 2, 4- thiazolidinedione, without Metformin, tin in our real world setting achieved greater absolute glycemic response had little effect on glucose lowering. In conclusion, Met-TZ showed excel- as compared to the controlled clinical trials, possibly as the higher baseline lent antidiabetic effects in an animal model of type 2 diabetes. If this for- HbA1c level at initiation is a strong predictor of HbA1c reduction. mulation proves to have reduced gastrointestinal effects compared to Met- Table. Demographics and Baseline Characteristics of Patients Initiated on formin alone, it might represent therapeutic advancement in the treatment Teneligliptin. Therapeutics of diabetes.

Clinical Diabetes/ Demographics/ Mean ± SD Range 95% Confi dence Interval p value

PUBLISHED ONLY Baseline Characteristics 2287-PUB Age (years) 50.5±11.21 24-74 47.9-53.1 < 0.0001 WITHDRAWN Baseline HbA1c (%) 9.5 ± 2.1 5.6-14 9-10 < 0.0001 BMI (kg/m2) 28 ± 4.4 17-39 27-29 < 0.0001 Duration of Diabetes (years) 7.2± 5.8 0.5-25 5.9-8.6 < 0.0001 2288-PUB Effect of Vildagliptin on Glycemic Control and Body Weight in Patients with T2DM Aged < 45 Years: A Pooled Analysis 2290-PUB NEELESH DONGRE, JAMES E. FOLEY, Basel, Switzerland, East Hanover, NJ WITHDRAWN Although correlation analysis indicates that the effect of vildagliptin (vilda) on HbA1c and weight is independent of age, such analysis can miss important differences in small under-represented age groups. Previously, we reported that in patients aged ≥75 years (yrs), there was no difference 2291-PUB in HbA1c reduction from baseline (BL) and in weight, vs. overall popula- Effects of Salsalate on Fasting Plasma Glucose and Cardiometa- tion (mean age ~55 yrs). Here we aim to compare effi cacy of vilda between bolic Parameters in Adults with Prediabetes: A Systematic Review younger (<45 yrs) and overall population. and Meta-analysis Data of patients aged <45 yrs with HbA1c and weight assessments were ALLISON R. WALCZYK, CARALYN M. DIGANGI, STEPHANIE C. GARCIA, ARLENE pooled from vilda clinical trials database. Changes (Δ) in HbA1c, FPG and SMALDONE, New York, NY weight from BL to week 24 were analyzed in drug-naive (DN) patients and While the role of insulin resistance (IR) and infl ammation in type 2 diabetes those receiving vilda/Metformin combination therapy (MC). (T2DM) is known, use of pharmacologic agents to mitigate progression from 2 The analysis included 439 patients: mean age 38.7 yrs; BMI 33.1 kg/m ; prediabetes to T2DM is an emerging area of research. The objective of this men 54%. DN (n=284) treated with vilda 50 mg qd/bid had BL diabetes dura- systematic review (SR) and meta-analysis (MA) was to evaluate the effects tion of 1.7 yrs and HOMA-IR score 7.6 while MC treated with vilda 50 mg bid of salsalate on fasting plasma glucose (FPG), insulin resistance (HOMA-IR), (n=155) had BL diabetes duration 4.2 yrs and HOMA-IR score 5.7. Changes low-density lipoproteins (LDL), and triglycerides (TG) in adults with predi- in mean HbA1c and FPG in both the DN and MC treated patients were simi- abetes. We searched 5 databases (PubMed, Embase, Cumulative Index to lar to overall population; weight neutrality was seen in overall population Nursing and Allied Health Literature, Cochrane Library, and New York Acad- and patients aged <45 yrs, despite improved metabolic control (Table). Thus emy of Medicine for randomized trials (RCT) comparing salsalate to placebo. there were no notable differences in the effect of vilda on HbA1c, FPG and Studies reporting study outcomes as mean±variance pre and post salsalate weight in younger patients vs. overall population. or change in means±variance over time were included in MA using random Table. Effi cacy Parameters from Baseline for Younger (<45 Years) and Over- effects models. Data were analyzed using Comprehensive Meta-Analysis all Population on Vilda 50 mg qd/bid and in Combination with Metformin at statistical software. Of 8 RCTs identifi ed (323 subjects, age 18-75, BMI >25 Week 24. kg/m2), 4 were included in MA. Study lengths ranged from 1-12 weeks. Tin- nitus was reported as an adverse event in 15 of 170 subjects (8.8%) who received salsalate. Compared to controls, FPG signifi cantly decreased after salsalate administration (pooled effect -1.36 [95% CI -1.89, -0.82]; 4 stud- ies 135 subjects) with no differences in HOMA-IR (pooled effect -0.38 [95% CI -1.01, 0.24]; 3 studies 98 subjects), and LDL (pooled effect -0.43 [95% CI -1.22, 0.35]; 3 studies 96 subjects). Salsalate’s short term effect on FPG in adults with prediabetes is promising; its effect in decreasing cardiovascular risk is less certain. Future studies of longer duration are needed to determine optimal salsalate dosing, longer term effect on glycemia and cardiovascular health, and potential adverse effects. Supported By: Novartis

For author disclosure information, see page A696. ADA-Supported Research

A578 CLINICAL THERAPEUTICS/NEW TECHNOLOGY—PHARMACOLOGICCATEGORY TREATMENT OF COMPLICATIONS

2292-PUB CLINICAL THERAPEUTICS/NEW TECHNOLOGY— Antidiabetic Effects of GMC-252, a Conjugate of Difl unisal and PHARMACOLOGIC TREATMENT OF COMPLICATIONS N-Acetylcysteine, in Combination with Metformin in db/db Mice SILVIA GARCÍA VICENTE, LUC MARTÍ, PIETER H. JOUBERT, IGNACIO FAUS, 2294-PUB MARTA SERRANO MUÑOZ, DAVID PÉREZ CÁCERES, ANTONIO ZORZANO, Esplu- gues de Llobregat, Spain, Barcelona, Spain WITHDRAWN Dietary control and Metformin still remain cornerstones of oral treatment of type 2 diabetes mellitus (T2DM). In spite of many second and third line options, there remains a need for oral therapies that do not cause hypo- glycaemia or weight gain. The GMC-252-L-Lysine Salt (GMC-252), a novel 2295-PUB molecular ester conjugate of difl unisal (targeting infl ammation) and N-Ace- Ameliorates Progression of Liver Tumors and Damage tylcysteine (targeting oxidative stress), is a novel antidiabetic approach in through Inhibition of Oxidative Stress, Infl ammation, and Fibrosis the early stages of drug development. We previously showed that GMC- in Nonalcoholic Steatohepatitis-Hepatocellular Carcinoma Mouse 252 (0.5 mmol/kg per day) reduced blood glucose, improved glucose toler- Model ance, and increased pancreatic insulin content without inducing weight gain M.S.T. REJINA AFRIN, SOMASUNDARAM ARUMUGAM, M.D. AZIZUR RAHMAN, in db/db mice. We now present further non-clinical data on the interaction VIGNESHWARAN PITCHAIMANI, VENGADESHPRABHU KARUPPAGOUNDER, between a lower dose of GMC-252 and Metformin. REMYA SREEDHAR, MEILEI HARIMA, HIROSHI SUZUKI, SHIZUKA MIYASHITA, 6-week old db/db mice (9-11) received daily oral doses of vehicle, a com- KENJI SUZUKI, HIROYUKI YONEYAMA, KAZUYUKI UENO, KENICHI WATANABE, bination of 0.2 mmol/kg of GMC-252 with 1.2 mmol/kg of Metformin, GMC- Niigata, Japan, Tokyo, Japan 252 alone or Metformin alone for 4 weeks. The dose of Metformin was Hepatocellular carcinoma (HCC) is the third most common cause of can- increased to 1.8 mmol/kg half-way through study in both groups receiving cer-related deaths, nonalcoholic steatohepatitis (NASH) is pivotal to link Metformin, due to the loss of measurable effects of the drug. The combina- diabetes with HCC. We investigated the effect of curcumin on the liver tion of GMC-252 and Metformin produced a signifi cant decrease in blood of novel NASH-HCC mouse model through modulation of oxidative stress, glucose, and an increase in pancreatic insulin, compared to GMC-252 or infl ammation, and fi brosis. Neonatal C57BL/6 male mice were exposed to Metformin given separately. The combination also enhanced the blood glu- low-dose streptozotocin (STZ) and were fed high-fat diet (HFD) at the age cose lowering effect of insulin action during an insulin tolerance test com- of 4 weeks and continued for 10 weeks, curcumin was given at 100 mg/ pared to the drugs given alone. kg dose by oral gavage daily after 10 weeks of STZ injection and contin- Therapeutics Our fi ndings, if confi rmed in humans, suggest that GMC 252 alone and in ued for 4 weeks along with HFD feeding. Steatohepatitis was characterized Clinical Diabetes/ PUBLISHED ONLY combination, with Metformin might bring a new option for oral regimens for by increased levels of hepatic pro-infl ammatory cytokines IP10, IFNγ, IL-1β, treating T2DM without hypoglycaemic risk and weight gain. TLR4 expression, HMGB1 translocation and NASH histological score. Our results show curcumin treatment improved the histopathological changes and signifi cantly reduced all these important infl ammatory markers in the 2293-PUB liver of NASH mouse. Curcumin treatment also signifi cantly reduced the Metformin Reduces Serum CA199 Levels in Type 2 Diabetes Chinese serum aminotransferases, and blood glucose levels Patients with Time-Effect and Gender Difference compared to NASH-HCC control group. Curcumin treatment downregulated WOLIN HOU, DANDAN ZHANG, FANG LIU, JUN YIN, WEI LU, MING LI, TAISHAN the hepatic upregulation of fi brosis (CTGF), oxidative stress (CYP2E1, C/ ZHENG, FENGDI LU, WEIPING JIA, Shanghai, China EBPβ, p67phox) and activation of NF-κB pathway in this mouse model. Sig- Background: This study was designed to clarify the infl uence of Metformin nifi cant elevation of VEGF, p-IRE1α, and p-PERK expression were found in on CA199 levels and its associated factors in Chinese T2DM patients. the NASH-HCC liver, which were signifi cantly attenuated by curcumin treat- Subjects and Methods: In total, 1,253 T2DM patients were enrolled, ment. Macrophage infi ltration and tumor formation in the liver of NASH- including a non-Metformin group (n = 616), a short-term Metformin group HCC mice were markedly inhibited by curcumin treatment. Finally, our results (n = 325), and a long-term Metformin group (≥ 2 years; n = 312). Their clini- indicate that curcumin may have the potential role to protect the liver in this cal and biochemical characteristics were collected and compared. After 1 novel NASH-HCC model. year, the biochemical parameters were re-examined in 296 patients. Sex hormones were determined, and associations between CA199 and other variables were assessed. 2296-PUB Results: At baseline, the incidence of abnormal CA199 levels was 14.7%, Effects of Exenatide on Fat Redistribution in Centrally Obese 8.9%, and 4.7% in the non-Metformin, short-term Metformin, and long-term Patients with Type 2 Diabetes Mellitus Metformin groups, respectively. CA199 levels in females were signifi cantly JIAQI CHEN, XINGCHUN WANG, LE BU, SHEN QU, Shanghai, China higher than in males and decreased signifi cantly with the time of taking Met- Glucagon-like peptide-1 receptor agonists (GLP-1 RAs) have been found to formin. The correlation and multiple stepwise regression analysis revealed induce weight loss in addition to improving glucose homeostasis. However, that glycosylated hemoglobin, Metformin, gender, total cholesterol, and fol- the effects of GLP-1 RAs on fat redistribution in centrally obese patients licle-stimulating hormone were independent impact factors on CA199 con- with type 2 diabetes mellitus have been poorly investigated. In a prospec- centrations. Binary logistic regression revealed that the risk of abnormal tive study, 38 centrally obese patients with type 2 diabetes were treated CA199 concentrations of the total population with short-term Metformin or with exenatide (5µg twice daily for 4 weeks increasing thereafter to 10µg long-term Metformin treatment decreased 11% (odds ratio = 0.89; P = 0.001) twice daily) for a total of 18 weeks. The patients’ glucose and insulin levels, and 30% (odds ratio = 0.70; P = 0.000), respectively, at baseline. After a and their lipid profi les were measured before and after exenatide treatment. 1-year follow-up, the incidence of high CA199 level decreased in both the In addition, homeostatic model assessments of β-cell function and insulin short-term and the long-term Metformin group compared with that of con- resistance (HOMA-B and HOMA-IR, respectively) were also calculated, and trols. The extent of CA199 decrease in the long-term Metformin group was the android fat mass, android/gynoid (A/G) ratio, and visceral adipose tissue the greatest, and the group’s risk of high blood CA199 level was reduced (VAT) area were measured by dual-energy X-ray absorptiometry (DXA). Of 67% (odds ratio = 0.33; P = 0.023). The reduction in women was more appar- the 38 patients who underwent baseline screening, 33 completed the study. ent than that in men (-18% vs. - 5%, P= 0.000). Exenatide treatment induced signifi cant decreases in body mass index (BMI) Conclusions: Metformin therapy reduced the CA199 level in Chinese and glycated hemoglobin (HbA1c). In addition, there were also signifi cant T2DM patients, and its greatest decrease occurred in women with longer improvements in β-cell function and insulin resistance, and the android fat therapeutic time. mass, A/G ratio, and VAT area were all signifi cantly decreased. Changes in Supported By: National Natural Science Foundation of China the A/G ratio were signifi cantly correlated with the changes in HbA1c and HOMA-B, while changes in the VAT area were signifi cantly correlated with the changes in HOMA-IR. Thus, exenatide treatment can induce a redistri- bution of adipose tissues, which may possibly contribute to better glucose homeostasis and an improved insulin response in obese patients with type 2 diabetes. (NCT02118376). Supported By: AstraZeneca China

ADA-Supported Research For author disclosure information, see page A696.

A579 HEALTH CARE CATEGORYDELIVERY—ECONOMICS

2297-PUB fi ltration rate (r=-0.64, r=-0.64, and r=-0.64, respectively). Additionally ApN Baseline Triglyceride Level Correlates with HbA1c Control by Vilda- correlated with high density lipoprotein, body mass index (BMI) and apolipo- gliptin Add-on to Metformin Treatment in Chinese Patients with protein A1 (r=0.37, r=-0.40 and r=0.44, respectively). Mean values for high- Type 2 Diabetes: A Subgroup Analysis of VISION Study sensitivity C-reactive protein (hs-CRP) at the beginning of the study were LI-NONG JI, LINGLI ZHOU, HUI FANG, LE WANG, BIN-HUI WANG, Beijing, China, 3.86±3.64, 2.67±2.52 and 5.31±2.37 in the Gr 1, 2, 3 respectively, and were Tangshan, China signifi cantly reduced by 0.63 (95% CI: 0.1-1.15), 1.35 (95% CI: -0.26-2.97) and It has been reported that obesity and increased non-esterifi ed fatty acid 1.71 (95% CI: 0.57-2.84) on average in all groups, with greater reduction in may impaired treatment effectiveness of GLP-1 or DPP-4 inhibitor through Gr 3 in comparison with Gr 2. HbA1c mean values were 8.01±0.79, 7.36±0.87 down-regulating of GLP-1 receptor signaling in animal model. Combined lipid and 8.01±0.95 in the Gr 1, 2, 3 respectively, and were signifi cantly reduced control medicine can improve the effi cacy of incretin-based therapies of dia- by 0.94 (95% CI: 0.73-1.15), 0.69 (95% CI: 0.05-1.32) and 1.15 (95% CI: 0.35- betes. To further explore the effect of lipid metabolism on diabetes treat- 1.95) on average, with no difference between groups. ADMA and BMI mean ment, subgroup analysis was performed using data from VISION, a multi- values in Gr 3 were 0.56±0.14 and 39.3±4.5 and were signifi cantly reduced center randomized controlled trial to study effi cacy of vildagliptin add-on to by 0.04 (95% CI: 0.02-0.06) and 2.65 (95% CI: 1.35-3.94) on average, whereas Metformin treatment in Chinese T2DM patients with inadequate glucose SBP reduction was signifi cant from baseline (137.5±16.9) in Gr 2 by 9.0 (95% control. Multivariate analysis showed that in vildagliptin add-on treatment CI: -0.05-18.55). Postprandial C-peptide and triglycerides (TG) were reduced group, baseline triglyceride (TG) levels was independent predictor for HbA1c in Gr 3 by -0.32 (95% CI: -0.65-0.01) and 0.67 (95% CI: -0.09-1.43) on average, reduction at both 12 weeks (p<0.001) and 24 weeks (p=0.05). When sub- however, not signifi cantly. Liraglutide was superior in hs-CRP, ADMA, BMI jects were stratifi ed, vildagliptin + low dose Metformin treatment showed and TG reduction in comparison with studied dipeptidyl peptidase-4 inhibi- signifi cant better effi cacy in subjects with lower baseline TG level <= 1.28 tors, therefore it is plausible to argue its advantage in diabetic angiopathy than those >2.03 mmol/L. In contrast, HbA1c reduction in Metformin titration protection. monotherapy group was not correlated with baseline TG. The results indi- cate baseline TG level infl uence the effect of vildagliptin, but not Metformin, may be through interaction at lipid metabolism level, . In clinical practice, HEALTH CARE DELIVERY—ECONOMICS individualized decision should be taken into account for selection of vilda- gliptin add-on to Metformin treatment. Further clinical trials are in need to 2300-PUB test the effi cacy of combination lipid control medicine with GLP-1 or DPP-4 Therapeutics inhibitor based diabetes treatments. (NCT01541956). WITHDRAWN Clinical Diabetes/ PUBLISHED ONLY 2298-PUB Ef fi cacy of Calcitriol in Treatment of Diabetic Leukoaraiosis Patients 2301-PUB with Mild Cognitive Impairment Risk for 30-Day Readmission in Diabetes XIAO-HUI ZHAO, YU-PING ZHU, JUAN YANG, HUI-QIN LIU, Shanghai, China ABEER ANABTAWI, YI ZHONG, JIANGHUA HE, DAVID ROBBINS, LELAND Aim: To assess the effect of calcitriol for diabetic leukoaraiosis patients GRAVES, Kansas City, KS with mild cognitive impairment (MCI). Reducing hospital readmissions, particularly within 30 days of discharge, Methods: A total of 256 patients with diabetic leukoaraiosis were con- has become a high priority for hospitals in light of decreasing payments by secutively enrolled in this study, of which 181 cases were diagnosed as MCI, Medicare and Medicaid (CMMS) for 30 day readmissions. To best plan inter- and were divided into two groups: control group (n=85), and intervention vention strategies, a retrospective analysis was performed at our tertiary group (n=88). Patients in intervention group received Calcitriol (0.25ug/day) care hospital including patients with a primary discharge diagnosis of diabe- treatment for 12 months. Cognitive function was assessed using Montreal tes (ICD-9 250.xx) between October 2011 and June 2014. Patients who had cognitive assessment (MoCA) scale at baseline, 6 months’ and 12 months’ a diabetes related readmission within 30 days of discharge from their index follow-up. admission were identifi ed and multiple characteristics were compared to Results: A number of 181 cases were randomly divided into control group those who were not readmitted including age, gender, marital status, ethnic- and intervention group. After 12 months, 173 cases (85 in control group, 88 ity, body mass index (BMI), type of diabetes, hemoglobin A1c (HbA1c), renal in intervention group) completed the follow up. In intervention group, MoCA function, length of stay, discharge disposition, and type of insurance. Data is scores were not signifi cantly improved at 6M ((18.84±2.08 vs. 19.23±2.42, presented as mean ± standard deviation or number (percentage). Two sample NS), but were signifi cantly increased at 12 month (18.84±2.08 vs. 22.32±2.36, t-test and Chi square test were used as applicable. A total of 646 patients P<0.01). Also, the scores were higher in intervention group than those in con- had an index diabetes related admission (355 males, 291 females) with mean trol group after 12 months’ follow up (22.32±2.36 vs. 20.84±2.37 P<0.01). age of 54.2 ±15.95 years. Seventy four patients (11.5%) were readmitted Language scores in intervention group at 6M (1.37±0.86), 12M (2.52±0.86) within 30 days following discharge. In comparison to those who did not need were signifi cantly higher than those at baseline (1.84±0.73), (P<0.01). And re-admission; patients who were readmitted within 30 days were younger they were also higher than those in control group (2.37±0.86 vs. 1.91±0.72 (48.3 ± 14.9 vs. 54.2 ± 15.9 years, p 0.0028), had lower BMI (Kg/m2) (28.4 ± at 6M, and 2.52±0.86 vs. 1.95±0.77 at 12M). After 12 months’ treatment, 6.8 vs. 31.2 ±8.2, p 0.0043), longer index hospitalization (days) (6.5 ± 5.16 vs. the scores of delayed memory increased from 1.30±1.41 to 2.77±1.84, signifi - 5.12 ± 4.92, p 0.0281), were more likely to have type 1 diabetes (p 0.0039), cantly higher than those at baseline and 6M (P < 0.01), whereas the scores Medicare insurance (p 0.0045), and less likely to have commercial insurance. at 6M did not differ from those at baseline. At 12M, there was signifi cantly No signifi cant difference was noticed with HbA1c, admission and discharge difference in the delayed memory scores between two groups (2.77±1.84 vs. serum glucose and creatinine, admission mortality risk, and discharge dispo- 1.41±1.46, P<0.01). No signifi cantly changes in in abstract function, visual sition. The 11.5% 30 day readmission rate was slightly below the reported space and executive function were observed in two groups (all P > 0.05). average of 14.4-22.7%. Diagnosis of type 1 diabetes, prolonged index admis- Conclusions: Calcitriol could improve the cognitive function of patients sion, lower BMI and Medicare coverage were associated with higher 30 day with diabetic leukoaraiosis-related MCI, among which the language ability readmission rate. Incorporating these factors into strategies to reduce read- improved remarkably. mission may be helpful in improving readmission rates.

2299-PUB 2302-PUB Liraglutide’s Impact on Infl ammation Might Be Superior in Angio- A Multidisciplinary Approach to Inpatient Diabetes Care Involv- protection Compared with Linagliptin and Vildagliptin ing Pharmacist Improves Glycemic Control and Prevents Hospital SPOMENKA LJUBIC, ANAMARIJA JAZBEC, ANTE PILJAC, MARTINA TOMIC, Readmission in High Risk Diabetic Patients MARIJANA VUCIC LOVRENCIC, Zagreb, Croatia KAI XIN NG, GOLDA Z.Z. WANG, ERIC KHOO, SHIH LING KAO, LIP MIN SOH, Sin- We compared the impact of incretins on infl ammation and diabetic angiop- gapore, Singapore athy. A total of 287 type 2 diabetics divided into three groups treated with Diabetes mellitus (DM) is common in hospitalized patients, with one study linagliptin (Group [Gr] 1; n=117), vildagliptin (Gr 2; n=102) and liraglutide (Gr 3; reporting that 19% of diabetic patients are readmitted within 1 month of dis- n=68) were studied during a 4-month follow-up period. Differences between charge for multiple reasons possibly relating to poor DM control. Overseas baseline and follow up values were analysed by t paired test. The signifi - pharmacist-led glycemic control team has been shown to improve glycemic cance level was considered at 5%. Asymmetric dimethylarginine (ADMA), control and reduced post-discharge hospital readmissions, however, there adiponectin (ApN) and homocysteine correlated signifi cantly with glomerular are currently no comprehensive Singapore data.

For author disclosure information, see page A696. ADA-Supported Research

A580 HEALTH CARE CATEGORYDELIVERY—ECONOMICS

A tertiary hospital in Singapore initiated a multidisciplinary care model performance of this basic CDS vs. an enhanced CDS system (with prioritized involving pharmacists in managing high risk diabetic patients, from the inpa- CDS to both patient and provider), we group randomized 6561 adults with tient through to the outpatient setting. Phone consult was introduced for diabetes to receive basic vs. enhanced CDS. Over a median of 14 months of selected individuals during the transition period to aid adjustment to their follow-up, Framingham 10-year absolute risk of fatal or nonfatal heart attack DM regimen. Readmission rates and glycated haemoglobin (HbA1c) 1 and or stroke declined from 27.8% to 24.8%, and Framingham reversible CV risk 3 months post-discharge respectively were collected for 160 patients. A (net of age and sex) declined from 12.6% to 9.6%. The proportion of subjects signifi cant reduction in HbA1c of 2.3% from baseline, with no readmission achieving goals increased for BP (40.7% to 78.3%), LDL (55.8% to 71.1%), and noted within 1 month post-discharge. Phone consult, which was provided to A1c (68.4% to 72.4%), and smoking declined from 29.4% to 23.4%. We con- 54% of patients, contributed to an additional 0.4-0.5% reduction in HbA1c clude that enhanced CDS was not superior to basic CDS, but that both CDS (p>0.05). A multidisciplinary approach to inpatient DM care involving phar- systems were associated with sustained clinically and statistically (p<.05) macist is effective in improving glycemic control and preventing hospital signifi cant reduction in CV risk in diabetes patients. readmission. However more robust data is required to better assess the Supported By: R01HL102144, P30DK092924 impact from the phone consult sessions. Table. HbA1c Reduction for Patients under Phone Consult and Those Without 2305-PUB Phone Consult. Partnering with Primary Care to Redesign and Improve Diabetes Total Phone Patient Pre-intervention Reduction in Care N= 160 Consult number HbA1c% HbA1c% KARLA K. GIESE, Albuquerque, NM (%) (Mean ± SD) (Mean ± SD) Larger health systems are called upon to provide leadership in improv- Patients reviewed 72 Yes 56 (77.8) 10.9% ± 2.3 -2.6% ± 2.3 ing population health and care redesign. A large, ethnically-diverse primary by DM pharmacists care practice in the southwest U.S. redesigned diabetes care after acquisi- No 16 (22.2) 9.8% ± 2.5 -2.1% ± 2.6 tion by a health system. Moving toward primary care medical homes, the poorly controlled diabetes subpopulation, previously referred to endocrinol- Patients not reviewed 88 Yes 30 (35.1) 10.8% ± 2.7 -2.3% ± 2.6 ogy outside the practice, was targeted for improvement. Care was success- by DM pharmacists fully redesigned with establishment of an onsite diabetes specialty clinic No 58 (65.9) 9.8% ± 2.4 -1.9% ± 2.6 for easy, effi cient referrals utilizing a nurse practitioner (NP) and a regis- tered dietician, both CDEs with established expertise in diabetes education, management and support. Medically complex patients or those with A1c > Therapeutics Clinical Diabetes/

2303-PUB 9% were referred to the NP, CDE. A retrospective convenience sample of PUBLISHED ONLY Implications of Using the UKPDS HbA1c Equation in Economic Mod- all new patients entering the NP’s care January-March the fi rst two years eling of Type 2 Diabetes Mellitus (T2DM) was analyzed (n=102). Hispanics accounted for 47%, mirroring the state’s MICHAEL WILLIS, CHRISTIAN ASSEBURG, PIERRE JOHANSEN, MELANIE SCHROE- census data of 47.7%. The difference between mean admission A1c (9.65%) DER, CHERYL NESLUSAN, Lund, Sweden, High Wycombe, United Kingdom, Raritan, NJ and 6 month follow up (7.30%) was statistically signifi cant (p ɖ 0.01) (Wil- Economic modeling is used to help allocate scarce health care resources coxon Matched Signed-Rank Test). A1cs of ɖ 7.5% were achieved by 77.4%, effi ciently. Because treatments have limited durability and T2DM is chronic while 59% achieved A1cs of ɖ 7.0%. Outliers of interest included a 70 year and progressive, assumptions about upward HbA1c “drift” over time are old referred with an A1c of 5.2% on basal bolus insulin along with Met- important. Most simulations use 1 of 2 approaches: linear drift in segments formin and a sulfonylurea experiencing symptomatic hypoglycemia. In con- or the non-linear UKPDS 68 evolution equation (both interrupted periodically trast, a 58 year old’s admission A1c was 17% post hospitalization for diabetic with rescue medication and corresponding HbA1c lowering). Because UKPDS ketoacidosis off insulin due to lack of insurance/medical care. Data collec- subjects received rescue medication, HbA1c-lowering effects are embedded tion coincided with roll out of the Affordable Care Act, which added new in the equation and explicit HbA1c lowering from modeled rescue medication patients to the practice of very low socioeconomic status, most previously constitutes double counting. We aim to raise awareness of HbA1c evolution uninsured. Clinical care factors such scheduling processes, qualitative data modeling by comparing model-based predictions with observed HbA1c tra- related to patient barriers to care, and future directions around returning jectories for the UKPDS and ADOPT trials. We used the validated Economic controlled patients to PCPs are also discussed. and Health Outcomes Model of T2DM to predict the UKPDS over 15 years and ADOPT over 5 years using 3 approaches for modeling HbA1c evolution: (1) 2306-PUB linear drift in segments, (2) UKPDS HbA1c equation with explicit rescue med- Diabetes INSIDE: A Quality Improvement (QI) Initiative to Address ication (potentially double-counting), and (3) UKPDS HbA1c equation without High A1c Values and Adherence to ADA Standards of Care in Pri- explicit rescue medication. Concordance of results with observed HbA1c tra- mary Care jectories was assessed. The fi rst approach matched the slope and between- KATHERINE L. NASHATKER, NOEL O. SANTINI, KELLIE RODRIGUEZ, UMA GUN- arm differences of the UKPDS well over the fi rst 13 years, though not pla- ASEKARAN, ELIZABETH E. OBIALO, SENTAYEHU KASSA, MANISHA RAJA, ROY teauing at the end as observed in the trial. Both UKPDS equation simulations E. FURMAN, CLAY V. TOWNSEND, ANNIE C. MATTHEW, STEVEN M. BOATRIGHT, matched the slope and plateau of the intensive arm well, but not the HbA1c LUIGI F. MENEGHINI, Dallas, TX, Media, PA time path of the conventional arm, and were associated with between-arm In collaboration with the ADA and a partner expert in QI we undertook an convergence not observed in the trial. The linear segment approach closely initiative to address patients with A1c > 9% (or missing A1c values) receiving matched the trajectories of HbA1c for each of the ADOPT treatment arms. care in the primary care clinics of a large urban safety net health system. The UKPDS HbA1c equation time paths resembled the observed HbA1c evo- From our diabetes registry we identifi ed patients seen ≥3 times within 18 lution for , but failed to predict the fl atter slopes for Metformin months. We formed a multidisciplinary committee with representation from and rosiglitazone. The linear segment approach can better replicate HbA1c all clinical sites and collectively agreed to focus on Hispanic patients with evolution as it is not confounded by the rescue medication effects embed- elevated A1c values through shared medical appointments (SMA), focusing ded in the UKPDS data. on insulin initiation or management. We have standardized the process for Supported By: Janssen Global Services, LLC identifying, referring, contacting, and scheduling patients into the appropri- ate SMA session and are deploying a standard approach to content, delivery, 2304-PUB and documentation of these visits. Comparative Effectiveness of Two Diabetes Clinical Decision Sup- The baseline demographics are shown below. Over half of patients with port Systems on Cardiometabolic Outcomes and CV Risk: Random- A1c > 9% were not on insulin therapy. At 6 months following implementation ized Trial of QI we anticipate a lower % of patients with elevated A1c values, as well PATRICK J. O’CONNOR, JOANN M. SPERL-HILLEN, A.L. CRAIN, KAREN L. MAR- as greater adherence to other ADA standards of care. GOLIS, WILLIAM A. RUSH, JAY R. DESAI, HEIDI L. EKSTROM, Minneapolis, MN A system-wide QI initiative utilizing an SMA-based approach is an effec- CMS and others are encouraging medical groups to reduce CV risk in dia- tive and adaptable strategy to address diabetes care in the primary care betes patients and other high risk adults. Electronic health record (EHR)- network of an urban safety net hospital. linked point of care Clinical Decision Support systems (CDS) can support this effort, and basic EHR-linked CDS that provides patient-specifi c treatment recommendations has previously been shown to signifi cantly improve glu- cose and some aspects of BP control in diabetes patients. To compare the

ADA-Supported Research For author disclosure information, see page A696.

A581 HEALTH CARE CATEGORYDELIVERY—ECONOMICS

Table. and urine microalbumin/Cr ratio (28% vs. 24%, p=NS). Average GMV patient All patients All with A1c>9% Hispanics A1c>9% satisfaction was 4.74 (Likert scale 1=worst, 5=best). Routinely, 2.7-fold more patients are seen per hour of endocrinologist time compared to individual N (%) 12,406 (100%) 2,874 (23%) 1,745 (14%) visits. We estimate that at optimal staffi ng effi ciency, the GMV will operate at Hispanic N (%) 6,917 (56%) n/a 1,745 (25%) $932 less per 4 hour session compared to traditional endocrine visits. Mean Age 56.6±0.1 53.2±0.2 52.3±0.3 An interdisciplinary specialty GMV model improves access to care at Gender 37±1% M 38±2% M 36±2% M lower cost for an underserved population. GMVs are a cost effective, viable alternative to traditional medical appointments. 2 Mean BMI (kg/m ) 33.2±0.1 33.2±0.1 32.40.2 Supported By: Nicholson Foundation Mean A1c (%) 7.9±0.0 10.8±0.0 10.7±0.0 % on insulin 37±0.0% 58±0.0% 55±0.0% 2309-PUB Mean LDL (mg/dl) 91.7±0.3 99.0±0.7 97.4±0.8 Designing Outpatient Diabetes Care Clinical Decision Support Sys- Mean S creat (mg/dl) 0.9±0.0 0.8±0.0 0.7±0.0 tems for High Use Rates PATRICK J. O’CONNOR, JOANN M. SPERL-HILLEN, JAY R. DESAI, KAREN L. MAR- % fl u vaccinations 83±1% 80±1% 84±2% GOLIS, Minneapolis, MN Supported By: Eli Lilly and Company; Novo Nordisk Inc.; Sanofi -Aventis To identify key design features of point-of-care diabetes clinical decision support (CDS) that have high use rates, high provider satisfaction rates, 2307-PUB and have improved control of major CV risk factors. Based on a series of Lean Problem Solving Methodologies Used to Improve Outpatient NIH-funded projects to develop point-of-care Electronic Health Record Diabetes Education Referrals from Inpatient Hospitalists: A Devel- (EHR)-linked Web based clinical decision support systems, we have iden- opmental Research Study tifi ed design features that contribute to observed high use rates (60-80%) LISA RAYNELLE SHELLEY, Hutto, TX at targeted visits, and high primary care provider (PCP) satisfaction rates Continuity of care is hallmark in healthcare, but there is a disconnect in trans- (94-95%), and positive effects on glucose and blood pressure control in lating inpatient diabetes care into outpatient diabetes programs (ODP). Our adults with diabetes. Key features of successful outpatient chronic disease goal is to generate ODP referrals in 25% (currently 18.5%) of newly diagnosed care clinical decision support system include the following: (a) co-designed by PCPs and researchers, (b) supported by clinic and medical group leaders,

Therapeutics diabetes inpatients or those with an A1c of 7.1 or higher. Increased referrals (c) designed to improve publicly-reported quality measures, (d) introspec- Clinical Diabetes/ should improve community health, facilitate continuity of care, and maintain

PUBLISHED ONLY recognition status as a diabetes education center. In order to address these tive identifi cation of targeted encounters, (e) target patients with potential issues and to see patients in a timely manner, we focused on group classes for substantive clinical benefi t, (f) have rooming nurse launch CDS early in with the capacity for 10 patients vs. individual sessions. To identify factors encounter workfl ow, (g) have PCP see CDS early in workfl ow for visit plan- impeding the referral process, we utilized the Lean problem solving methodol- ning, (h) have patient review CDS before PCP enters room, (i) simple visual ogy. We assessed the number of patient referrals from the inpatient facility to display of potential benefi ts for patients, (j) prioritization of treatment the ODP between June 1, 2013-October 31, 2014. Of the 973 patients admitted, options based on potential benefi t to patient, (k) ongoing feedback of CDS zero were referred. The fi ndings were that hospitalists desired ODP referrals use rates to PCP and clinics, (l) compensate clinics to cover training costs, for their patients, however, the discharging provider did not enter the order. (m) locate algorithms in Web service to facilitate updates and scalability, (n) Referrals improved when inpatient dietitians, who see all patients who are build in EHR order sets to facilitate clinical actions. These CDS design and newly diagnosed or have an A1c of greater than 7.1, placed orders in the elec- implementation features are generally associated with high use rates, high tronic medical record for the ODP pending physicians signature. This method PCP satisfaction rates, and clinical improvement. However, tailoring the fea- increased the inpatient referrals from 0 to 58 per year, exceeding our goal. This tures to particular practice settings is necessary. resulted in improved patient continuity of care and access to diabetes edu- Supported By: DK068314, HL102144, DK092924 cation. There was also an increased awareness of ODP availability resulting in improved diabetes self-management. This study shows that the utilization 2310-PUB of inpatient dietitians to initiate referrals for ODP increases the actual refer- PAD: A New Model to Care Complications in Type 2 Diabetic rals and results in improved continuity of care. Future research should focus Patients on the translation of inpatient referrals to outpatient scheduling, incidence of OLGA E. DISOTEO, EMANUELE SPREAFICO, ELENA CIMINO, GIANLUIGI PIZZI, readmission and emergency room visits, as well as improving the number of Milan, Italy patients who complete all four sessions of the group diabetes classes. In our hospital we have planned a new model to prevent, care and moni- tor balance and complications in type 2 diabetic patients, called Diagnostic 2308-PUB Outpatient Pathways (PAD-Percorso Ambulatoriale Diagnostico). Aim of this Specialty Diabetes Group Medical Visits Improve Access to Care observational study is to evaluate feasibility, effi cacy and diffi culties of these and Clinical Outcomes at Lower Cost in an Underserved Patient new pathways in the biggest Lombardy Hospital after a test of two years. Population From October 2013 to November 2015, we have performed 414 PAD, involved VALERIE S. GANETSKY, RACHEL ADAMS, TYLER COOLMAN, JEFFREY BRENNER, 207 patients, 125 M, 82 F, median age 67 ys (30-87). PAD consists in blood STEVEN KAUFMAN, Philadelphia, PA, Camden, NJ tests (HbA1c, total cholesterol, HDL chol, triglycerides, microalbuminuria, renal The Cooper University Hospital Urban Health Institute in Camden, New and liver function), ECG, eye examination, carotid dopplerultrasound, second Jersey offers interdisciplinary, specialty diabetes group medical visits level evaluations, if necessary, and fi nally diabetologic visit. Every patient per- (GMV) to medically underserved patients in an urban, resource-poor com- formed one PAD per year and another diabetologic visit to 6 months. The 207 munity. The GMV was created to address the long lag times for endocrine pts have been followed in this model for 57 weeks ± SD 27, 52. appointments within a resource-constrained environment. We studied the At the fi rst PAD we observed HbA1c mean 60,8 mmol/mol ± SD 15,2, 137 effects of the GMV on access to care, clinical measures, and cost. pts with one or more complications and good glycemic control, 25 pts with The GMV schedules up to 13 patients per hour and is composed of a pre- HbA1c > 64 without complications, 45 pts with complications and HbA1c > 64. visit chart review, questionnaire completion by a “Navigator” (APN, LPN, At the second PAD we observed HbA1c mean 58,6 mmol/mol ± 16,1, 142 endocrine fellow, or PharmD), endocrinologist medical review, and facili- pts with complications and good glycemic control, 20 pts with HbA1c > 64, tated group discussion. Access to care and clinical diabetes measures were 40 pts with complications and HbA1c < 64, 5 patients without complications collected and compared between GMVs and individual urban endocrine and with good glycemic control. No patient lost to follow up or developed visits. Patient satisfaction was surveyed. Comparative salaries of staff were new complications and the glycemic control is stable in the group without used for cost analysis. important change in therapy. During PAD two patients have been hospital- A total of 272 patients accounted for 703 GMV visits from 12/1/14-11/30/15. ized, one for anemia in intestinal cancer and one for bradyarhythmia. The majority of patients were Hispanic (58%) or African-American (35%), and Our considerations about PAD are positive, this model can improve adher- 46% had Medicaid insurance. Time to next new patient appointment avail- ence to specialistic visit and complication screening in force of hospital cen- ability was 14 days in the GMV and 28 days in the individual urban endocrine tralized planning, and also the compliance to the therapy can improve for a group (p<0.001). Mean baseline HbA1c in both groups was 9.3%, with a 0.23% motivational reinforcement due to the involvement in the patient care of greater reduction in the GMV group (p=NS). GMV patients had higher rates of more professional roles. The problems are waiting lists, defi ciency and orga- up-to-date HbA1c (77% vs. 65%, p<0.05), lipid panel (66% vs. 53%, p<0.01), nization of staff, lack of fund for an emerging widespread disease.

For author disclosure information, see page A696. ADA-Supported Research

A582 HEALTH CARE CATEGORYDELIVERY—ECONOMICS

2311-PUB months with/without other antidiabetes medications (ADM) seen as part of Real-Time Monitoring Telehealth Solution for Patients with Diabetes their usual care in 18 countries. Data were measured during patient visits to MARIT L. BOVBJERG, JENNEY LEE, ROSA WOLFF, MICHAEL A. MAY, Corvallis, OR physicians semiannually for 2 years. Insulin progression was defi ned as an Intensive outpatient care for patients with complex chronic conditions will increase in antidiabetes treatment complexity. Glycemic control goal at Year be an essential part of our evolving healthcare system. We created an app- 2 was defi ned as HbA1c ɖ7%. Multiple imputation via chained equations based telehealth platform to deliver Intensive Case Management (ICM) and was used for missing data. During the study period, a total of 924 patients health coaching to patients with diabetes. Realtime monitoring is facilitated had insulin progression (34.1%). Of those, only 28% achieved glycemic con- by a cellular, data-enabled glucometer. We are currently using the app in 2 trol at Year 2. Those who achieved glycemic control were older, with lower patient populations. Group 1 is high-cost, high-utilizing, Medicaid patients baseline HbA1c; majorities were males, non-smokers, and with family his- of a multispecialty clinic in the Pacifi c Northwest. A team of nurses rotates tory of diabetes. Additionally, 30% of the group who achieved glycemic con- responsibility for remote monitoring, education, trouble shooting, and other trol (ɖ7%) still had an HbA1c level above their physician’s stated treatment ICM-related patient interactions. Group 2 is a national healthcare person- goal. In conclusion, there are important demographic and clinical differences nel company, whose diabetic employees interact on the app with health at baseline among T2D patients with insulin progression by glycemic control coaches; all clinical inquiries are referred to the primary care home. Overall status. in Group 2, there was no signifi cant change in blood glucose (ordinary least Table. Characteristics of Patients Who Progressed Insulin. squares regression, clustering by individual, p = 0.7) over the fi rst 6 months Baseline Characteristic* HbA1c ɖ7% HbA1c >7% P of the intervention. However, Group 2 individuals are stably employed and (n=255) (n=669) value** well educated; many of them began with well-controlled diabetes. Among Group 2 individuals whose mean glucose readings were >150 mg/dL at Age, years 62.7±10.6 59.9±10.5 0.01 baseline, more than half improved during the fi rst 6 months; average read- Male 135 (53) 333 (50) 0.45 ings dropped by up to 23.7 mg/dL. In Group 1, >60% of patients improved HbA1c, % 7.4±1.8 8.5±1.8 <0.01 during the fi rst 6 months; the overall mean glucose fell by 0.36 mg/dL/week Family history of diabetes 144 (56) 408 (61) <0.01 (p < 0.05). We also assessed patient and provider satisfaction, other patient- centered health outcomes, and change in healthcare costs. Results will be ADM Basal insulin 122 (48) 330 (49) 0.76 compared between the two groups; we will also present results of predic- Short-acting insulin 27 (10) 64 (10) 0.75 tive models determining characteristics of patients most likely to benefi t Mixed insulin 65 (25) 179 (27) 0.72 from this app-based ICM approach. Metformin 114 (45) 374 (56) <0.01 Therapeutics Supported By: InterCommunity Health Network Sulfonylurea 62 (24) 199 (30) 0.14 Clinical Diabetes/ Dipeptidyl peptidase-4 inhibitor 28 (11) 77 (11) 0.79 PUBLISHED ONLY Glucagon-like peptide-1 agonist 11 (4) 25 (4) 0.66 2312-PUB Thiazolidinedione 9 (4) 40 (6) 0.20 Could Glycated Hemoglobin Be Equally Useful as Control Assess- ment Method in Different Therapeutic Groups? Region Asia 87 (34) 248 (37) 0.45 TZVETELINA TOTOMIROVA, IVONA DASKALOVA, Sofi a, Bulgaria Europe 72 (28) 124 (19) <0.01 Glycated hemoglobin (HbA1c) is mainly used for assessment of long-term Middle East 42 (17) 123 (18) 0.61 glycemic control in patients with type 1 and type 2 diabetes, refl ecting a North America 31 (12) 91 (14) 0.64 period of previous 8-12 weeks. But is there a dependence on therapeutic Latin America 23 (9) 83 (12) 0.30 regimen? For these purpose we studied 106 patients (60 men, 46 women; * Mean±SD for continuous and No (%) for categorical variables, ** Univariate mean age 58.23±10.81 years) with type 2 diabetes (31 on oral therapy group linear or logistic regression. 1; 33 on pre-mixed insulin, group 2; 21 on multiple insulin injections, group Supported By: Eli Lilly and Company 3) and type 1 (21 patients, group 4). Patients were asked to perform multiple daily blood glucose measurements (eight point profi le at least three days per week) of fasting and prandial blood glucose for three month period. HbA1c 2314-PUB was measured at the end of this period. Correlation between HbA1c and Building Diabetes Champions in Primary Care: The Air Force Expe- blood glucose measurements in different day time was estimated. In group rience 1-coeffi cient of correlation with mean glucose was ρ=0.699, p<0.01, in DARRICK J. BECKMAN, MARK W. TRUE, JANA L. WARDIAN, CONNIE C. MOR- group 2-ρ=0.452, p<0.01, in multiple injection treated groups no correlation ROW, NINA A. WATSON, TOM J. SAUERWEIN, San Antonio, TX was found-for type 2 diabetes ρ=0.308, for type 1 diabetes ρ=0.111. In oral The Department of Defense (DoD) provides comprehensive care to over treated group and in patients on pre-mixed insulin, HbA1c showed good cor- 150K patients with diabetes (active duty, retirees, and dependents) in over relation with mean blood glucose before and after breakfast, before and 400 clinics worldwide. With only 33 endocrinologists in the DoD, the major- after lunch, before and after dinner, before bedtime and at night (p<0.05). ity of diabetes care is left to primary care. No such correlation was found in group 3 and group 4. There was not found The U.S. Air Force Diabetes Center of Excellence designed the Diabetes correlation with glucose variation in neither of the groups. In patients with Champion Course (DCC), a semi-annual, 3-day course, to train primary care type 2 diabetes receiving oral treatment or pre-mixed insulin, HbA1c is valu- teams to become local champions in diabetes standards of care. Didactics able control assessment method. In cases treated with multiple daily insulin and hands-on training are focused on current standards, the latest technolo- injections (type 1 and type 2 diabetes), HbA1c is not strongly correlated with gies, and patient fl ow in a team-based setting. Each team is tasked to iden- blood glucose measurements and should be used for control assessment tify local defi cits and make a Plan of Action (POA) for implementation. only in combination with self-monitoring of blood glucose. HbA1c is not suf- In a process improvement project, we evaluated the impact of this course. fi cient to refl ect glucose fl uctuations in each of therapeutic group. We contacted attendees from 43 clinics with POAs from the last 3 courses, and we received 25 responses from 19 clinics. Results were obtained via a semi-structured phone interview to evaluate benefi ts of and obstacles to 2313-PUB POA implementation. Demographic and Clinical Factors May Predict Insulin Progression The last 3 courses had 222 participants from 77 clinics: 47 physicians, 23 and Glycemic Control among Patients with Type 2 Diabetes Using mid-level providers, 113 nurses, and 39 from other disciplines. Twelve of 19 Insulin: Learning from the MOSAIc Study clinics (63%) were able to implement part or all of their original POAs and AYAD K. ALI, BRADLEY H. CURTIS, WESLEY EDDINGS, JAMES R. ROGERS, 17 of 19 clinics (89%) reported improvements associated with attending the MENGDONG HE, ABDURRAHMAN ABDURROB, BRUNO LINETZKY, JAY BAE, course. Benefi ts included improving multidisciplinary cooperation, routine RAN DUAN, STEVEN BAIN, MANOJ CHAWLA, YOEL TOLEDANO, IKURO MAT- screenings (labs, foot exams, ophthalmology), and documentation; setting SUBA, MUSTAFA ARAZ, WILLIAM H. POLONSKY, SEOYOUNG C. KIM, Indianapo- up a diabetes education class; and increasing collaboration between the 3 lis, IN, Boston, MA, Buenos Aires, Argentina, Swansea, United Kingdom, Mumbai, military services. Common obstacles to POA implementation were turnover India, Petah Tikva, Israel, Kanagawa, Japan, Gaziantep, Turkey, Del Mar, CA of staff, lack of time and resources, provider resistance, and lack of leader- MOSAIc is a multinational 2-year prospective cohort study designed to ship support. understand challenges associated with insulin progression in patients with The DCC provides the expert-driven guidance and tools to primary care type 2 diabetes (T2D). This analysis describes characteristics of patients clinical teams to increase adherence to diabetes standards and appropri- who had insulin progression in the study period according to glycemic con- ately leverages the limited number of endocrinologists in the DoD. The trol status. Eligible patients were ≥18 year old, T2D, using insulin for ≥3

ADA-Supported Research For author disclosure information, see page A696.

A583 PEDIATRICS—OBESITYCATEGORY AND TYPE 2 DIABETES

DCC will continue to be modifi ed to address common barriers to maximize Table. impact. Total Obese T2D p (n=103) (No T2D) (n=13) values 2315-PUB (n=90) LPN-led Insulin Titration Protocols Improve Hemoglobin A1c Levels Age (years) 12.8 (3.2) 12.9 (3.3) 12.4 (2.2) 0.25 in an Urban, Underserved Primary Care Offi ce % Female 58% (60) 57% (51) 69% (9) 0.39 RACHEL ADAMS, VALERIE GANETSKY, TYLER COOLMAN, JEFFREY BRENNER, STEVEN KAUFMAN, Camden, NJ, Philadelphia, PA % High Risk Ethnicity 83% (85) 82% (74) 85% (11) 0.83 The Cooper Health System Urban Health Institute in Camden, New Jersey Family History of T2D on 47% (48) 43% (39) 69% (9) 0.08 offers an LPN-led insulin protocol to a high-risk, urban, resource-poor Maternal Side patient population with diabetes. Long wait times for primary care appoint- Family History of T2D on 20% (21) 19% (17) 31% (4) 0.32 ments and specialty endocrine care limit access to medication adjustment Paternal Side and detract physicians from more complex medical decision making. Family History of T2D on 9% (9) 8% (7) 15% (2) 0.36 Insulin titration protocols for basal, premixed, and basal/bolus insulin regi- Maternal and Paternal Sides mens developed by an endocrinologist, APN, and PharmD were used by LPNs BMI z Score 2.2 (0.4) 2.2 (0.4) 2.3 (0.3) 0.18 to promote patient education and improve glycemic control. Pre- and post- % Acanthosis Present 81% (83) 78% (70) 100% (13) 0.06 intervention HbA1c readings were collected and assessed. Patients with an HbA1c > 7.0% and on insulin were referred from the university primary care HOMA-IR x 4.3 (3.2) x x clinic once insulin was initiated or if the patient had an elevated HbA1c. Hemoglobin A1c (%) 6.1 (1.6) 5.7 (0.4) 9.3 (2.8) 0.0003 Patients were enrolled in 4 sessions for blood glucose review, education, and insulin titration. The education curriculum covered blood glucose log review, insulin injection technique, hyper- and hypoglycemia recognition and 2318-PUB management, dietary education using The Plate Method and diabetes self- Triglycerides/Glucose Index Is a Useful Surrogate Marker of Insulin care behaviors using videos and other instructional materials. LPNs received Resistance in Adolescents 20 minutes of weekly, diabetes-focused education from an endocrinologist, BORAMI KANG, YEO-REE YANG, HAE KYUNG YANG, EUN YOUNG LEE, SUN- APN, or Pharma.

Therapeutics YOUNG LIM, JIN-HEE LEE, KUN-HO YOON, Seoul, Republic of Korea 37 patients were seen over 90 visits. Mean age was 59.5 yrs. The popula-

Clinical Diabetes/ Obesity induces insulin resistance (IR), which plays a major role of type 2

PUBLISHED ONLY tion was African American (24%) and Hispanic (76%), and 51.3% had Med- diabetes, cardiovascular disease and metabolic syndrome (MetS) in adoles- icaid insurance. Of the 21 patients eligible for a 12-week follow-up HbA1c, cents, as well as adults. Some previous studies have suggested that the Trig- 0%, 37.4%, 31.3% and 31.3% of the patients completed 1, 2, 3 or 4 sessions, lycerides/glucose Index (TyG index) as a surrogate marker of insulin resis- respectively. 16 of 21 eligible patients completed the pre- and post HbA1c tance in adult. The aim of this study was to evaluate the TyG index and its levels with a mean decrease of 1.89% (95% CI, -0.41 to -3.38, p< 0.05). association with IR according to weight categories and metabolic abnormal- Low resource primary care provider offi ces have limited resources for ities in adolescents. This study was performed from May to June 2014 as a patient education. Task shifting insulin titration from physicians to LPNs cross-sectional research in Chung-ju City, Korea. A total of 222 nondiabetic enables relationship building, increases patient education, and frees physi- subjects fourth grade of elementary school or fi rst year of middle school cian appointments for complex decision-making in a cost effi cient manner. were included in this study. Obese and overweight were defi ned as BMI Supported By: Nicholson Foundation (kg/m2) by sex and age were over 95%ile and 85%ile respectively. HOMA-IR and TyG index were calculated using the following formulae, respectively: 2316-PUB [fasting insulin (mIU/L) xFPG (mmol/L)]/22.5, ln [triglycerides (mg/dL) xFPG (mg/dL)/2]. Metabolic syndrome (MetS) was defi ned according to IDF crite- WITHDRAWN ria in adolescents. The mean value of the TyG index was higher in the group with MetS (n=6) compared to the group without MetS (n=216) (8.75±0.24 vs. 8.07±0.42; p<0.001), and increased according to the number of meta- bolic abnormalities (0 (n=166): 7.98±0.35, 1 (n=34): 8.310±0.48, 2 (n=16): PEDIATRICS—OBESITY AND TYPE 2 DIABETES 8.54±0.43 and 3 (n=6): 8.75±0.24; p < 0.001). TyG index was well correlated with HOMA-IR (R=0.435, p<0.001). There were statistically signifi cant dif- ferences of HOMA-IR (p<0.001) and TyG index (p<0.001) among the three 2317-PUB weight categorical groups (normal=165, overweight=31, obese=26). The Which Obese Youth Are at Increased Risk for Type 2 Diabetes? Post-hoc comparison result showed that HOMA-IR in obese group was only LINDSEY WALDMAN, BRITTANY WISE, MARINA GOLDIS, ELIZABETH BURT- higher than other groups and there was no statistically signifi cance between MAN, MOLLY REGELMANN, ROBERT RAPAPORT, New York, NY normal and overweight group. Whereas, TyG index showed statistically dif- The American Diabetes Association recommends screening children for ference between normal and overweight group. Our results suggest that TyG type 2 diabetes (T2D) who are overweight with 2 or more risk factors: family index can be a useful discriminant marker of MetS and a early surrogate history (FH) of T2D, high risk ethnicity, signs of or conditions associated with marker of IR in adolescents. insulin resistance and maternal history of diabetes (DM) or gestational DM. Referring all patients meeting screening criteria leads to heavy burden and expense. We aim to determine differences between obese patients who 2319-PUB were and were not diagnosed with T2D at referral to our DM clinic. Long-Term Course of a Japanese Patient with Leprechaunism: Use Data were obtained by retrospective chart review of continuous consulta- of Recombinant Metreleptin tions of obese youth at risk of T2D between 5 and 20 years at the Division SHUNTARO MORIKAWA, TAKESHI YAMAGUCHI, KATSURA ISHIZU, TOSHIHIRO of Pediatric Endocrinology and Diabetes at the Icahn School of Medicine TAJIMA, Sapporo, Japan, Shimotsuke, Japan at Mount Sinai between December 2013 and 2015. Information obtained Leprechaunism (Donohue Syndrome) is a rare autosomal recessive dis- included age, ethnicity, T2D in fi rst or second degree relatives, body mass ease that is characterized by severe insulin resistance caused by muta- index z-score, presence of acanthosis, hemoglobin A1c (HA1c) and homeo- tions of . Because of the poor prognosis, there is little report stasis model of assessment of insulin resistance (HOMA-IR) by fasting blood about long-term course of leprechaunism patient. Here we report the clinical glucose and insulin. Cases were excluded if data were incomplete. Statistics course of a Japanese patient with leprechaunism over twenty years and the included t Test, Chi square and Mann Whitney U tests (MWU). experience of recombinant metreleptin induction. Our data show there is no signifi cant difference in factors other than HA1c The patient is now a 26 years old Japanese female. She was born as (MWU) between patients that were and were not diagnosed with T2D. While a small for gestational age and her phenotypic symptoms (fl accid skin, we were not able to identify statistically signifi cant risk factors between the exophthalmos, low-set-year and big mouth) and hyperglycemia accom- 2 groups, we feel that our data may have been limited by a small sample size panied by hyperinsulinism were noticed. Molecular analysis of INSR gene and referral bias. revealed a missense mutation (p.P87L) in the paternal allele and a 1.3 kb deletion between exons 4 and 6 in the maternal allele. Based on these fi nd- ings, she was diagnosed as leperchaunism and rh-IGF1 subcutaneous injec- tion had been started since 6 months old. The effi ciency of rh-IGF1 gradually

For author disclosure information, see page A696. ADA-Supported Research

A584 PEDIATRICS—TYPECATEGORY 1 DIABETES decreased and her control of diabetes worsened. She showed neovascular onstrated that the longer patients remained in the transition clinic, the glaucoma and retinal neovascularization and her vision lost at 12 years old. lower the HbA1c over time (p = 0.008). GEE model also indicated that those She also developed endometrial cancer at 21 years old. At 24 years old, she with private insurance had lower HbA1c levels before, at, and after transi- was introduced recombinant metreleptin subcutaneous injection therapy tion (p=0.043). Both older age and being Caucasian were related to lower (started from 0.03mg/kg/day and continued in 0.10mg/kg/day). Metreleptin HbA1c levels, but were of borderline signifi cance. These results suggest led to the loss of her appetite, body weight and body fat. However, it was that the time in transition clinic is important for improved glycemic con- diffi cult to control her blood glucose levels. Thus, metreleptin therapy was trol among DM1 patients. We will study factors that contribute to declin- stopped and high-dose insulin was restarted. ing trends in HbA1c, including psychological and physical factors, so that It has been reported that metreleptin therapy is effective for Rabson- interventions to improve HbA1c levels can be proposed. At the same time, Mendenhall syndrome. However, metreleptin therapy did exert no benefi cial we will continue to recruit participants, increasing the sample size to raise effect in our case. This difference may be due to the severity of the disease. statistical power. Leprechaunism is still intractable, and thus it is necessary to develop new Supported By: Harold Hamm Diabetes Center treatment. 2322-PUB Transition from Pediatric to Adult Care for Youth with Type 1 Diabe- PEDIATRICS—TYPE 1 DIABETES tes: A Systematic Review and Meta-analysis ALAN T. SCHULTZ, ARLENE SMALDONE, New York, NY 2320-PUB Adolescents with type 1 diabetes (T1DM) struggle to maintain glycemic Factors Associated with the Reduction of Psychological Distress in control. An additional challenge is the transition from pediatric to adult care. Parents of Children Diagnosed with Type 1 Diabetes The American Diabetes Association has called for more research on struc- EINAS ALKHATIB, VIRGINIA UHLEY, BERNARD DEGNAN, Rochester, MI, Royal tured transition. The purpose of this systematic review (SR)/meta-analysis Oak, MI (MA) was to examine which structured transition programs are effective in This research aimed to explore parents’ perception of distress regard- improving outcomes following transition. Following PRISMA guidelines, we ing his/her child’s diagnosis of type 1 diabetes, as well as determine which searched 6 databases (CINAHL, Cochrane, Embase, Medline, PsychINFO, factor (s) parents identify as being most effective in reducing distress. Exist- PubMed) for studies that assessed the effi cacy of a structured transition ing studies often focus on child physical health, rather than parent psycho- program on A1c. Studies reporting A1c or change in A1c for the intervention Therapeutics logical health. vs. control group pre and post transition were pooled using a random effects Clinical Diabetes/

Surveys were designed to inquire which components of diabetes distress MA model. Of 2552 studies identifi ed, 17 (1 randomized control trial [RCT], 6 PUBLISHED ONLY were experienced, as well as rate stress reduction factors. Similar to the quasi experimental, 1 prospective and 9 retrospective cohort) met inclusion Diabetes Distress Scale, factors were rated from 1-6 (1 being not applicable criteria. Findings represent data from 3,282 youth with T1DM (51.7% male, or least effective at stress reduction, and 6 being most effective). During age 16-22 years) undergoing transition. Programs varied and included transi- pediatric endocrinology appointment check-in at Beaumont, 100 surveys tion coordinators (n= 7), transition clinics with pediatric and adult providers were distributed to parents of children who have had type 1 diabetes for at (n= 3), transition clinic (n=9), and group education meetings (n=5). Average least one month. Statistical analysis included logistic regression and odds duration of transition programs was 4 years. Transition interventions had no ratio to determine which factors were most effective. effect on A1c at 12 months (mean change -0.12 [95% CI -0.33-0.08]; 3 studies Logistic regression with a biostatistician was used to analyze the 81 sur- representing data from 392 youth). Of other outcomes studied (clinic atten- veys completed. At least one diabetes distress symptom was experienced by dance [n=11], severe hypoglycemia [n=6], diabetic ketoacidosis [DKA] [n=7], 93% of parents. The symptoms and percentages experienced were: stress and diabetes-related hospitalizations [n=5)], transition programs showed (88%), frustration (77%), anxiety (71%), burn out (45%), and concern (39%). greatest consistency in reducing DKA (4 of 7 studies) and diabetes related Several factors were most effective in stress reduction: endocrinology hospitalizations (4 of 5 studies). SR fi ndings suggest that transition inter- appointments (4.95/6), glucose monitor and A1c levels (4.43/6), increased ventions may reduce DKA episodes and diabetes-related hospitalizations time since diagnosis (4.31/6), health insurance (4.16/6), and increased age or but not improve A1c. Further research is needed to determine which pro- independence of child (4.03/6). Additional factors rated include brochures, gram types are most effective for adolescents with T1DM during transition childcare, diabetes camp, dietician, electronic applications, family member/ to adult care. friend who is a healthcare professional, family member/friend whose child has type 1 diabetes, insulin aspart (NovoLog), insulin lispro (Humalog), insu- 2323-PUB lin pump, Juvenile Diabetes Research Foundation, sports, school nurse, Novel Education and Diabetes Care Pilot Program for Latinos Using school teacher, and support groups. Group-Setting Appointments (GSA) To conclude, endocrinologists can use these results to further provide ANDREA GERARD GONZALEZ, Aurora, CO effective stress coping methods to parents of children diagnosed with Latinos with type 1 diabetes (TID) face challenges related to language and type 1 diabetes. cultural barriers resulting in high risk of poor diabetes control, infrequent use of technology, frustration and fi nancial burden to the health system. We are 2321-PUB conducting a 3-year study of Spanish-speaking (SS) Latino patients using a HbA1c Levels after Entering a Diabetes Transition Program new model of GSA for education and care to determine if this will improve JAMES T. LANE, JONI K. BECK, CYNTHIA MOHAMEDAGIC, MEGAN SHORT, their diabetes outcomes in a cost-effective way. The program also fosters a JUSTIN DVORAK, YING ZHANG, Oklahoma City, OK strong community for this population. Study participants are T1D SS patients The OUMC College of Medicine has sponsored a transition program for seen at the Barbara Davis Center (BDC) aged 1-23 years. The study is divided young adults with diabetes for 4 years, open to patients transitioning from into a control group who receives routine care and patients who participate Pediatric Endocrinology (Peds Endo) to Adult Endocrinology. There remains in the Latino Program (LP). Patients in the LP receive routine care visits alter- concern that the transition process is diffi cult with less than acceptable nating with LPGSA in which educational materials are covered in a group levels of control exhibited, most recently by the T1D Exchange Clinical setting including activities and interactions focused on peer to peer learn- Registry in 2015. We sought to compare our patients with type 1 diabe- ing and sharing. The LP focuses on education in three main areas: nutri- tes (DM1) to the T1D Clinical Registry results. We recruited DM1 patients tion, behavior and medical management, including technology and research. that had previously consented to participate. Participants fi lled out annual Groups are scheduled by patient age ɖ12 years and >12 yrs. A multi-disci- health questionnaires, had annual labs, and took a variety of psychological plinary team of fl uent Spanish speaking providers assists in LP delivery. In tests. Patients were recruited while in Peds Endo to allow for observation the fi rst 6 months, 85 study participants (33 boys, 51 girls) have completed of changes across the transition. Of the 47 recruited, the mean age at tran- the fi rst LP GSA visit. Study participants had baseline mean HbA1c of 9.08%. sition was 19.5 ± 1.4 y (mean ± SD), while age at diagnosis was 10.2 ± 4.6 The median change in HbA1c in LP participants over 3 months was -0.05% y. Mean duration of diabetes was 9.6 ± 4.8 y. Most were Caucasian (82%), (p=0.45). 3 participants transitioned from multiple daily injections to insulin had completed high school, (89%), had private insurance (75%), and the pumps after participating in the LP. A total of 202 family members partici- majority were male (62%). HbA1c measurements were obtained before, at, pated in the in fi rst LP education visits. Satisfaction data from question- and after transitioning from Peds Endo to transition clinic. A moving aver- naires after the fi rst visit were highly positive. Preliminary results show that age graph showed HbA1c levels peaked around the transition visit, then study participants in the LP who return for their 3-month routine follow-up decreased thereafter. A general estimation equation (GEE) model dem- visits, have lower HbA1c trend, although not statistically signifi cant, and are

ADA-Supported Research For author disclosure information, see page A696.

A585 PEDIATRICS—TYPECATEGORY 1 DIABETES

returning for their second LP visit. We have seen 29 patients return for the Figure 2. second LP visit (phase 2 started Nov 2015). Supported By: The Leona M. and Harry B. Helmsley Charitable Trust

2324-PUB Transition Referral Practices for Young Adults with Type 1 Diabetes: National Survey of Pediatric Endocrinologists SHIVANI AGARWAL, KATHARINE C. GARVEY, JENNIFER RAYMOND, MARK SCHUTTA, Philadelphia, PA, Boston, MA, Denver, CO Transfer from pediatric to adult care for young adults (YA) with type 1 diabetes (T1D) is a vulnerable time marked by gaps in care and deteriora- tion of glycemic control. Standardized protocols for establishment of adult diabetes care are not available, and little is known about pediatric endo- crinologist (PE) referral patterns in the United States. To characterize PE experiences, referral practices, and barriers to transitioning YA with T1D, we fi elded a web-based survey to PEs in the American Medical Association Masterfi le. We received 142/1020 completed surveys (response rate 14%) representing 32 states. The majority of PEs (age 44±10; yrs in practice 12±11) were female (66%) and worked in academic centers (75%). Half endorsed patient age as the main reason for initiating transfer while only 13% had a transition policy. The most frequently endorsed barriers to timely YA trans- fer included reluctance to end a long therapeutic relationship (75%), lack of 2326-PUB guidance from a transition policy (46%), inability to identify an adult diabetes Association of Glycemic Control, Body Mass Index, and Race with provider (AD) (45%) and concern about inadequate resources in adult care Age at Menarche in Girls with Type 1 Diabetes: SEARCH for Diabe- (45%). Many (~40%) PEs felt that ADs lacked training in YA T1D care and that tes in Youth mental health access was most lacking in adult care (63%). To facilitate YA HERMANN BORG, WEI LANG, RALPH B. D’AGOSTINO, JEAN M. LAWRENCE, Therapeutics transfer, PEs desired lists of ADs experienced in YA T1D care (79%); use of CATHERINE PIHOKER, GRACE KIM, STEVEN L. YOUNG, PAUL WADWA, WILLIAM Clinical Diabetes/

PUBLISHED ONLY transition readiness tools (63%); and adoption of transition policies (54%). V. TAMBORLANE, ELIZABETH MAYER-DAVIS, Chapel Hill, NC, Winston-Salem, NC, Survey responses did not differ by PE gender, age, yrs in practice, or practice Pasadena, CA, Seattle, WA, Denver, CO, New Haven, CT setting (Fisher exact p≥0.05). Our results underscore key barriers to timely Delayed age at menarche (AAM) in girls with type 1 diabetes (T1D) has establishment of adult diabetes care for YA with T1D from the perspective been reported, yet its predictors remain elusive. Data from 401 girls (mean of referring PEs including PE emotional attachment to YA patients, lack of age 9.59 ± 2.69 yrs) diagnosed with T1D from 2002-2005 (T1D duration 9.75 ± transition policies, and perceived lack of AD T1D expertise and resources. 6.13 months) with a baseline visit prior to menarche and ≥1 follow-up visits, These fi ndings highlight potential targets for intervention such as provider were used to examine the impact of baseline glycemic control (mean A1c continuing medical education, and widespread dissemination of transition 7.76 ± 1.36%), BMIz (mean 18.54 ± 3.41) and race (78% non-Hispanic white) policy and preparation tools to PEs. on AAM. The unadjusted mean AAM for girls with poorly controlled T1D (A1c≥9.9%, n=28) was 12.9±2.1 yrs. There was a borderline difference 2325-PUB (p=0.0435) between the AAM of this group and AAM of 12.4±1.4 yrs of opti- Pancreatitis in Children with Type 1 Diabetes mally (A1c<7.5%, n=198) and AAM of 12.2±1.3 yrs of intermediately con- VICTOR S. HARRISON, Philadelphia, PA trolled (7.5%ɖA1c<9.9%, n=174) groups. The Figure shows the scatter plot Acute Pancreatitis (AP) and Diabetic Ketoacidosis (DKA) are known to co- of A1c vs. AAM. Signifi cant negative correlation between BMIz and AAM exist but the pathophysiology, prevalence and optimal diagnostic modality, was shown (p<0.001). There was no signifi cant difference in AAM by race. are unknown. We compiled a small retrospective case series of 8 patients Multiple linear regression models were fi tted to examine the impact of A1c, with type 1 diabetes (DM1) treated at the Children’s Hospital of Philadelphia BMIz and race on AAM. A1c when adjusted for socioeconomic status, BMIz who presented with and/or Lipase >3x the upper limit of normal; and T1D duration had signifi cant linear and non-linear association with either with and without coexisting DKA. AP was confi rmed by imaging in AAM (p<0.001). BMIz had signifi cant independent effect on AAM (β=-0.55; 2/20 episodes, and in all cases recovery was imminent. Like previous stud- p<0.0001). Race was not signifi cantly associated with AAM. In conclusion, ies, we show that lipasemia is less specifi c than amylasemia in defi ning AP glycemic control and BMIz have independent opposite effects on AAM in during DKA. We report for the fi rst time a linear relationship between lipase girls with T1D. and amylase in the setting of DKA in children. Furthermore, our fi nding of Figure. unexplained pancreatitis in children with DM1, without DKA suggests that AP may cause or worsen DKA. Clinicians should use amylase to rule out AP during DKA. Figure 1.

For author disclosure information, see page A696. ADA-Supported Research

A586 PEDIATRICS—TYPECATEGORY 1 DIABETES

2327-PUB 2329-PUB The Use of Continuous Glucose Monitors in Adolescents with Poorly Anti-aging Gene Klotho Attenuates Pancreatic β Cell Apoptosis in Controlled Type 1 Diabetes (T1D) on Multiple Daily Injection (MDI) Type 1 Diabetes Insulin Regimens YI LIN, ZHONGJIE SUN, Oklahoma City, OK CEDRIC K. NG, CLEMENT C. CHEUNG, JAMIE R. WOOD, Los Angeles, CA Apoptosis is the major cause of death of insulin-producing β-cells in type 1 Management of adolescents with poorly controlled T1D is clinically chal- diabetes mellitus (T1DM). Klotho is a recently discovered anti-aging gene. We lenging. The use of continuous glucose monitoring systems (CGM) has proven found that klotho gene is expressed in pancreatic β-cells in humans and mice. effective in optimizing diabetes care in those who are well-controlled. We The purpose of this study is to investigate if Klotho protects cells. Interestingly, conducted a randomized controlled pilot study using the DEXCOM G4 CGM halplodefi ciency of Klotho (KL+/-) exacerbated streptozotocin (STZ)-induced with share to assess whether its use would improve overall diabetes care diabetes (a model of T1DM), including hyperglycemia, glucose intolerance, in poorly controlled adolescents with T1D who are on basal-bolus MDI regi- diminished islet insulin storage, and increased apoptotic cells. Conversely, in mens. vivo β-cell-specifi c expression of mouse Klotho gene (mKL) attenuated β-cell Thirty patients aged 10-18 years old with T1D for at least one year and apoptosis and prevented STZ-induced diabetes. mKL promoted cell adhesion HgbA1c levels >9% were recruited for 6 months. All patients were on a to IV, increased FAK and Akt phosphorylation, and inhibited Caspase basal-bolus MDI insulin regimen. Mean age of subjects and diabetes dura- 3 cleavage in cultured MIN6 cells. mKL abolished STZ- and TNFα-induced inhi- tion was 15.6 ± 1.96 years and 5.3 years (4.33-9.83 years), respectively. bition of FAK and Akt phosphorylation, Caspase 3 cleavage, and cell apopto- All patients received standard of care treatment and monthly phone calls sis. These promoting effects of Klotho can be abolished by blocking between visits. The intervention group had 16 patients who wore a CGM β1. Therefore, these cell-based studies indicated that klotho protected cells device while the control group had 14 patients. by inhibiting β-cell apoptosis through activation of the integrin β1-FAK/Akt Twenty-fi ve patients showed up for their 3 month follow up visit (Control pathway leading to inhibition of Caspase 3 cleavage. In an autoimmune T1DM group = 13, CGM group = 12). Baseline HgbA1C levels were 11.0% in the con- model (NOD), we showed that in vivo cell-specifi c expression of mKL improved trol group and 10.71% in the CGM group. Preliminary data showed a decreas- glucose tolerance, attenuated cell apoptosis, enhanced insulin storage in cells, ing trend in HgbA1c at the 3 month visit, with net changes of -0.71% in the and increased plasma insulin levels. The benefi cial effect of Klotho gene deliv- CGM group and -0.72% in the control group. There was a trend in increased ery is likely due to attenuation of T cell infi ltration in pancreatic islets in NOD number of blood glucose (BG) checks per day at the 3 month follow up visit, mice. Overall, our results demonstrate for the fi rst time that Klotho protected with net changes of +1.18 in the CGM group and +0.74 in the control group. cells in T1DM via attenuating apoptosis. The difference between groups in HgbA1c and daily checks did not reach Supported By: National Institutes of Health (DK093403, HL118558, AG049780) Therapeutics statistical signifi cance. Clinical Diabetes/ PUBLISHED ONLY HgbA1c levels improved in both groups and there was also an increase 2330-PUB in average number of daily BG checks, although there was no statistically Factors Affecting Health Related Quality of Life (HRQoL) in Pediat- signifi cant difference in improvement when comparing the two groups at 3 ric Patients with Type 1 Diabetes Mellitus (T1DM) and Their Care- months. These improvements may have been due to increased contact by givers in Spain: Results from the CHRYSTAL Observational Study the diabetes team between visits. Whether these improvements are sus- MAGALY PEREZ-NIEVES, JUAN PEDRO LOPEZ SIGUERO, JULIO LOPEZ BASTIDA, tainable and whether a difference will be observed between groups at 6 JESUS REVIRIEGO, JUAN OLIVA MORENO, DINGFENG JIANG, RENATA VILLEGA, months remains to be seen. TATIANA DILLA, MARIA MERINO, ISAAC ARANDA-RENEO, LUIS A. VÁZQUEZ, Indianapolis, IN, Malaga, Spain, Madrid, Spain, Alcobendas, Spain 2328-PUB Despite the growing prevalence of T1DM in Spain and its consequences Adverse Effects of Overweight and Obesity on Pupillary Function in on health and social costs, little is known about the overall HRQoL of pedi- Type 1 Diabetes atric patients and their caregivers. The objective of this study was to assess NALINI M. SELVEINDRAN, PAUL Z. BENITEZ-AGUIRRE, MARIA E. CRAIG, ALBERT overall HRQoL for pediatric patients with T1DM and their caregivers using K. CHAN, JANINE CUSUMANO, KIM C. DONAGHUE, Sydney, Australia a generic preference-weighted instrument EuroQol 5 dimensions (EQ-5D). Identifying determinants of diabetic autonomic neuropathy (AN) is desir- The study also evaluated how glycemic control and diabetic complications able as it is a serious complication of type 1 diabetes (T1D). Pupillometry can or comorbidities (DCC) affect HRQoL in this population. CHRYSTAL obser- detect early changes in the autonomic nervous system. vational study was conducted in 2014 on a representative sample of 275 Aim: To study the longitudinal association between pupillometry param- patients aged 1-17 years, diagnosed with T1DM and distributed across 12 eters and risk factors for AN in 316 youth with T1D between 1990 and 2015 hospitals in Spain. Patient/caregiver pairs were stratifi ed by patient’s glyce- with at least two assessments and minimum follow-up 7 years. Phasic light mic control based on HbA1c level (HbA1c ≥7.5% vs. HbA1c 7.5%) and by the refl ex and resting pupillary diameter were assessed using an infrared com- presence or absence of DCC. The EQ-5D questionnaire was administered to puterized pupillometer. Longitudinal analysis was performed using genera- caregivers, who responded both as proxies for patients and for themselves. lised estimating equations (GEE). Baseline patient data: age 12.7 [11.4-14.3] Responses were converted into utility scores along a continuum extending years, mean (SD) HbA1c 8.4 (0.9)% and 29% overweight/obese. At last fol- from death (0.0) to full health (1.0) and are presented in Table 1. Glycemic low-up: age 24.3 [21.8-27.5] years, mean HbA1c 8.4 (0.7)% and 56% over- control and the presence of DCC may impact HRQoL of pediatric patients weight/obese. All pupillary function tests declined with longer diabetes with T1DM and their caregivers. Resource allocation decisions may benefi t duration (p<0.001). Explanatory variables in GEE models were: age, gender, from these results. diabetes duration, BMI SDS (or BMI), SBPSDS, DBPSDS, total cholesterol, Table 1. HRQoL Measured by EQ-5D for Overall Population and Stratifi ed by mean HbA1c, insulin dose/bodyweight and urine albumin excretion. BMI was Glycemic Control (HbA1c < 7.5 or ≥ 7.5) and Presence or Absence of DCC. signifi cant when modelled categorically (overweight/obese vs. normal, see Table) and as a continuous variable for refl ex amplitude and maximum con- striction velocity. Obesity, along with dyslipidaemia and hyperglycaemia are signifi cant modifi able risk factors for AN in our T1D population. Table. Multivariable GEE Model. Resting pupillary Refl ex Max Constriction diameter Amplitude Velocity β (95% CI) β (95% CI) β (95% CI) Duration (years) -0.04 (-0.05, -0.04) -0.03 (-0.03, -0.02) -0.06 (-0.08, -0.05) Overweight/Obese – -0.09 (-0.14, -0.04) -0.19 (-0.35, -0.04) Cholesterol (mmol/l) -0.07 (-0.12, -0.02) – – Mean HbA1c (%) -0.10 (-0.18, -0.02) – – Only signifi cant results, P<0.05 are reported in Table. Supported By: Eli Lilly and Company

ADA-Supported Research For author disclosure information, see page A696.

A587 PREGNANCY—BASIC SCIENCE/TRANSLATIONAL CATEGORY PREGNANCY—CLINICAL/EPIDEMIOLOGY

2331-PUB PREGNANCY—BASIC SCIENCE/TRANSLATIONAL Glucose Variability Predicts Infl ammation in Adolescent Type 1 Dia- betes 2333-PUB ROBERT P. HOFFMAN, AMANDA S. DYE, HONG HUANG, JOHN A. BAUER, Colum- bus, OH, Charleston, WV, Lexington, KY WITHDRAWN Adolescents with type 1 diabetes (T1D) have increased infl ammation and oxidation which lead to endothelial damage. It is unclear whether hyper- glycemia or glucose variation is more damaging. This study was designed to determine the relationship of infl ammatory (c-reactive protein, CRP), oxi- 2334-PUB dative (total anti-oxidative capacity, TAOC and malondialdehyde, MDA) and WITHDRAWN endothelial (soluble intracellular adhesion molecule 1, sICAM1) markers to glycemic control measures from continuous glucose monitoring (CGM, 3 d, Medtronic Guardian) and to HbA1c, and HbA1c x duration area under the curve (A1cDur). Seventeen adolescents (8 F; age, 13.1 ± 1.6 yr (mean±SD); 2335-PUB 2 duration, 4.8 ± 3.8 yr, BMI, 20.3 ± 3.1 kg/m ; A1c, 8.3 ± 1.2%) were studied. WITHDRAWN Log CRP tended to increase as duration increased (r=0.42, p=0.1) but was not related to age, BMI, HbA1c, or A1cDUR. TAOC increased as logA1cDUR increased (n=13, r=0.61, p=0.028). CRP and sICAM were not related to CGM average glucose but log CRP increased as 3 d glucose standard deviation 2336-PUB (GSD) increased (r=0.66, p=0.006). LogsICAM 1 (r=0.46, p=0.07) tended to do RANKL Levels Are Lower in Pregnant Women with GDM Compared the same. TAOC increased as GSD increased (r=0.63, p=0.028) (Figure). MDA with Pregnant Women without GDM was not related to CGM results. Increased glucose variability is associated DEIRDRE COCKS ESCHLER, GEORGIA KULINA, ADOLFO GARCIA-OCANA, ELINA with increased infl ammation and possibly increased endothelial damage in TROFIMOVKSY, KHADEEN C. CHEESMAN, RUPANGI VASAVADA, THOMAS adolescents withT1D. Increased TAOC with increasing variability may be an KRAUS, CAROL LEVY, New York, NY, Bridgeport, CT effort to compensate for the ongoing oxidative stress. There are both short and long term complications to the mother and the Figure. fetus in a pregnancy complicated by gestational diabetes mellitus (GDM). Therapeutics

Clinical Diabetes/ While the precise mechanisms in preserving glucose balance in a healthy

PUBLISHED ONLY pregnancy are unknown, various growth factors and hormones have been implicated or associated with GDM risk, including Vitamin D, OPG, and HGF, in humans or rodents. Prior studies have shown signifi cantly lower vitamin D levels in women (W) with GDM. RANKL, a cytokine thought to be pre- dominantly involved in bone metabolism, is also a potent NF-κB activator, a mediator known to be involved in hepatic insulin resistance and B cell dys- function. RANKL and its receptor, RANK, have been identifi ed in liver and pancreatic β tissue and elevated levels have been associated with cardio- Supported By: National Institutes of Health vascular disease and to be predictive of type 2 diabetes (T2DM) in humans. Blocking hepatic or systemic RANKL in mice with T2DM improved even nor- 2332-PUB malized, glucose levels. Compared to non-pregnant (PREG) controls, RANKL Provider Perceptions of Level of Risk Involved in Diabetes-Specifi c levels are higher in the second and third trimester of a healthy pregnancy Risk-Taking Behaviors and lower in PREG W with preeclampsia. PREG W with GDM are known to RACHEL M. WASSERMAN, YULIANA ROJAS, CAROLINE GONYNOR, BARBARA have a higher risk of preeclampsia. Prior research has shown a trend towards J. ANDERSON, DAVID D. SCHWARTZ, Houston, TX lower RANKL in PREG W with GDM compared to non GDM. We performed Because general risk-taking behavior (e.g., reckless driving; binge drink- a cross-sectional, IRB-approved study. Blood samples were collected from ing) peaks in adolescence, adolescents may also take risks with their diabe- consented PREG GDM (n=39) and non GDM (n=42) at the 24-32 week obstet- tes management. It is important to know which behaviors pose the highest rical visit when the 100g oral glucose tolerance test was completed. Sub- risk so that interventions may be targeted accordingly. We solicited pro- jects were matched for age, BMI, gestational age. Blood plasma was ana- vider perceptions of level of risk involved in diabetes-specifi c risk-taking lyzed for RANKL, OPG, PRL, TRAIL, HGF, PAI-1 and TNFa. RANKL levels were behaviors. Thirteen pediatric endocrine providers (MDs, NPs, and a CDE) lower in GDM subjects (p =0.019). RANKL/OPG ratio trended lower in GDM completed a 38-item questionnaire assessing “how likely is this behavior to but did not meet signifi cance (p=0.15). These results could suggest that there result in a serious health problem for an older teen/young adult with type 1 are alterations in bone metabolism in PREG W with GDM. A lower RANKL diabetes?” Answers were provided on a 5 point Likert scale from 1: Not Risky may be a marker or increase the risk for GDM. Further evaluation of vitamin to 5: Extremely Risky. Providers’ responses ranged from 1 to 5, with 4 (Very D and ambient glucose will be evaluated to assess whether these factors Risky) being the most common response (Avg M=3.8). On average, providers may have impacted results. rated the following items as most risky: youth went 24 hours without insulin (M=4.9), youth got drunk to the point where s/he could not take care of his/ her diabetes (M=4.8), youth used drugs when no around knew s/he had dia- PREGNANCY—CLINICAL/EPIDEMIOLOGY betes (M=4.7), youth participated in an organized sport without telling the coach s/he has diabetes (M=4.5), and youth felt BG might be low and did not 2337-PUB treat it (M=4.5). Responses to each item were generally consistent across Hemoglobin Glycation Index Predicts Macrosomia through Its Asso- providers (Avg SD=0.8). Providers disagreed most on the following: taken ciation with Mean Glucose in Pregnancies Complicated by Preex- less insulin than s/he knew s/he needed (SD=1.3, Range=1-5), taken pump isting Type 1 Diabetes off and did not reconnect it right afterwards (SD=1.2, Range=2-5), went to JAN SKUPIEN, IZABELA LASON, KATARZYNA CYGANEK, BARBARA KATRA, sleep after sex without checking BG (SD=1.2, Range=1-5). Providers viewed ALICJA HEBDA-SZYDLO, IZABELA JANAS, IWONA TRZNADEL-MORAWSKA, most items on this questionnaire as risky, indicating support for face valid- PRZEMYSLAW WITEK, ELZBIETA KOZEK, MACIEJ T. MALECKI, Krakow, Poland ity. Discrepancies could indicate different interpretations of the question Hemoglobin glycation index (HGI) is defi ned as a difference between or true disagreement. Providers may wish to assess diabetes-specifi c risk- measured glycated hemoglobin A1c (HbA1c) and HbA1c level predicted taking behaviors, particularly those that are believed to be most dangerous. from patient’s average glycemia. It has been designed as a marker captur- Future research is needed to determine the frequency of diabetes-specifi c ing between-individual propensity for glycation, but its use and interpreta- risk-behaviors and association with health outcomes in youth with type 1 tion has been controversial. It is unknown if in pregnancies complicated by diabetes. diabetes, HGI can be useful as an independent risk predictor, additional to Supported By: National Institutes of Health; National Institute of Diabetes and HbA1c. Here we assess HGI as a predictor of adverse pregnancy outcomes in Digestive and Kidney Diseases women with type 1 diabetes (T1D). We analyzed 442 singleton pregnancies in women diagnosed with T1D at least 1 year before conception. HbA1c was assayed in every trimester in 375 term pregnancies and in the fi rst two tri-

For author disclosure information, see page A696. ADA-Supported Research

A588 PREGNANCY—CLINICAL/EPIDEMIOLOGYCATEGORY mesters in 67 women with preterm delivery. Mean glucose in each trimester (HOMA-B), homeostasis model assessment insulin resistance index (HOMA- was calculated as an average of 7 to 9 daily glucometric self-measurements IR) were evaluated. over one week preceding the corresponding HbA1c measurement. We used Results: The levels of serum vitamin D in GDM group was signifi cantly linear regression to predict HbA1c from the mean glucose and calculated lower than that in control group [(10.16±1.45) vs. (15.22 ±3.20) ng/mL HGI as the difference between observed and predicted HbA1c. Using logistic (P<0.05)]. The correlation analysis showed that vitamin D level was positive regression models adjusted for same-trimester HbA1c we found that HGI in associated with HOMA-B (P<0.05), but was negative correlated with HOMA- the second trimester was an independent from HbA1c predictor of macroso- IR. Muitiple regression analysis showed that age, BMI and vitamin D level mia: higher observed then predicted HbA1c had protective association with were independent infl uencing factors of GDM. macrosomia (p=0.016, odds ratio for 1% HGI increase 0.33). The association Conclusion: Patients in GDM group had a lower Vitamin D level compared was attenuated in the third trimester (p=0.086, odds ratio 0.52). There was to healthy pregnant women, vitamin D defi ciency may affect the islet β-cell no signifi cant association of HGI with preterm deliveries, low birth weight, function and increase insulin resistance, thus contributing to the develop- stillbirths of congenital malformations, when adjusted for HbA1c. In con- ment of gestational diabetes. clusion, HGI is associated with the risk of macrosomia, but not with other Supported By: National Natural Science Foundation of China adverse pregnancy outcomes in T1D. Direction of association suggests that HGI captures changes in glycemic control not refl ected by HbA1c. Our data 2341-PUB do not support hypothesis that individual’s propensity for glycation is associ- Impact of Maternal Triglyceride and Novel Metabolic Parameters ated with adverse pregnancy outcomes in T1D. on Neonatal Anthropometrics in Asians LEE LING LIM, PREMILLA ARUL ARUMUGAM, NUR SULWANA MOHAMAD 2338-PUB HANAPI, LUQMAN IBRAHIM, SHARMILA SUNITA PARAMASIVAM, JEYA- IADPSG and ALAD Criteria for the Diagnosis of Gestational Diabe- KANTHA RATNASINGAM, PENG CHIONG TAN, SITI ZAWIAH OMAR, NOOR tes Mellitus (GDM): What Happens in the Gap? AZMI MAT ADENAN, AZANNA AHMAD KAMAR, MUHAMMAD YAZID JALALU- FABIOLA ROMERO, MARIA LIS ALARCON, LILIANA VIGO, ELIZABETH VALINOTTI, DIN, AZRIYANTI ANUAR ZAINI, NURSHADIA SAMINGAN, SIEW PHENG CHAN, LILIAN ZORRILLA, ALICIA GOMEZ, ZUMILDA IBARRA, NIDIA VILLALBA, CARMEN ALEXANDER TONG BOON TAN, SHIREENE RATNA VETHAKKAN, Kuala Lumpur, ECHAGUE, ALEJANDRA OCAMPO, ALBA GUERRERO, DIEGO VERA, HELEN LOPEZ, Malaysia NADIA GARCIA, FRANCISCO CABRERA, ROSA VILLALBA, SANTIAGO BAEZ, CAR- In Caucasian populations, maternal obesity and triglyceride (Tg) in addi- OLINA ACOSTA, MARIA CELIA MENONI, Asunción, Paraguay tion to glucose also impact neonatal anthropometrics via intrauterine fetal There are few studies of treatment of GDM with the International Associ- programming. Less is known of associations between these novel metabolic Therapeutics Clinical Diabetes/

ation of the Diabetes and Pregnancy Study Groups (IADPSG) criteria in Latino parameters and neonatal body composition in Asians, in whom strength PUBLISHED ONLY population, which mainly uses Association Latin-American Diabetes (ALAD) of/ thresholds for correlation may differ given that visceral adiposity is criteria. Determine whether the adoption of the IADPSG criteria involve an increased at a lower BMI in Asians. Hence we aim to determine the relation- increase of fetal maternal morbidity in patients with GDM by ALAD crite- ship between maternal BMI/metabolic variables (including Tg) and neonatal ria, but “missing” by IADPSG criteria, is unknown. We perform a prospec- anthropometrics in Asians. tive, longitudinal study of 896 consecutive cases from High Risk Obstetrics In this prospective observational study, women in Kuala Lumpur, Malay- Service of Hospital Central Institute Prevision Social, Asunción, Paraguay, sia were recruited at 14-32 weeks gestation and blood drawn for insulin, from January 2013 to December 2014, to assess neonatal and obstetric out- C-peptide and Tg during a 75g OGTT. Women with GDM and normal glucose comes among women who test positive for ALAD criteria but negative for tolerance (NGT) were enrolled, excluding pre-pregnancy diabetes. Neonatal the IADPSG criteria (untreated) compared with women who would test posi- anthropometrics were assessed. Neonatal fat mass (NFM) was calculated tive for the IADPSG criteria (treated) and normotolerance pregnant women. as per Catalano et al’s formula. According to OGTT 75 grams glucose: Group 1 IADPSG: Fasting plasma glu- We recruited 211 women: 33 GDM, 178 NGT-Malay 134 (63.5%), Chinese cose ≥ 92 mg/dl and or, 1 hour ≥ 180 mg/dl and or, 2 hours ≥ 153 mg/dl. Group 49 (23.2%), Indian 23 (10.9%) and others 5 (2.4%). In the whole group (n=211), 2 ALAD no IADPSG: Fasting plasma glucose ɖ 91 mg/dl and 1 hour ɖ 179 mg/ pregravid BMI correlated positively with BW/BW ratio [BWR]/NFM (r=0.189, dl and 2 hours 140-152 mg/dl. Group 3 Normotolerance: Fasting plasma glu- 0.277, 0.355 respectively; p<0.05). Fasting glucose (r=0.194, p<0.05) but not cose ɖ 91 mg/dl and 1 hour ɖ 179 mg/dl and 2 hours ɖ 139 mg/dl. They were Tg, correlated with BW. While 2-hour insulin correlated with BW (r= -0.159, in Group 1: 412 (46%), Group 2:59 (7%) and Group 3: 425 (47%) women. Age p<0.05), correlation between HOMA2-%S and BWR was weaker (r= -0.137, was 31 ± 6, 30 ± 5 and 28 ± 4 years respectively (p = 0.00023) There was no p=0.052). Amongst NGT mothers with pregravid BMI ≥27.5 kg/m2 (n=104), difference in weight of new born: 3050, 2950 and 2980 grams in Group 1, 2 both fasting glucose and Tg correlated signifi cantly with BWR (r=0.263, and 3 (p:0.16) respectively, nor in frequency of caesarean section, Group 1: 0.207; p<0.05) while HOMA2-%S correlated with BWR/NFM (r= -0.200, 72.6%, Group 2: 65.5%, Group 3: 73.5% (p : 0.51), nor in macrosomia or GEG -0.293; p<0.05). 12, 4%, 17%, 9% (p: 0.085), prematurity 30%, 32.8%, 31.4% (p : 0.84), and In conclusion, Asian NGT mothers with pregravid BMI ≥27.5 kg/m2 (Asian small for gestational age, 21.4%, 25.5%, 27.1% (p :0.15) respectively. In con- obesity cut-offs), Tg and insulin resistance predict BWR. Importantly, risk clusion, the group of women diagnosed with GDM with ALAD criteria that of fetal adiposity occurs at lower BMI thresholds compared to Caucasians. were not identifi ed by IADPSG criteria and therefore untreated, did not show Studies with larger population samples are warranted to confi rm these pre- a signifi cant increase of maternal and fetal complications with comparation liminary fi ndings which may indicate the need to target maternal Tg and insu- to the treatment group (IADPSG) and normotolerants women. lin resistance in pregnancy. Supported By: University of Malaya 2339-PUB WITHDRAWN 2342-PUB The Effi cacy and Safety of Continuous Subcutaneous Insulin Infu- sion vs. Multiple Daily Injection in Pregnant Women with Type 1 Diabetes Mellitus: Systematic Review and Meta-analysis 2340-PUB PRZEMYSLAW RYS, SLAWOMIR SIEJKA, PIOTR WOJECIECHOWSKI, AGNIESZKA Serum Vitamin D Levels in Gestational Diabetes Mellitus Patients LUDWIG-GALEZOWSKA, MACIEJ T. MALECKI, Krakow, Poland and Its Relationship with Islet Cell Function An adequate glycemic control in pregnant women with type 1 diabetes FANG LIU, WOLIN HOU, JUNLING TANG, JUNXI LU, RUI HE, WEIPING JIA, Shang- mellitus (T1DM) constitutes the most important part of diabetes manage- hai, China ment. Randomized clinical trials (RCTs) has shown the advantage of continu- Objective: To analyze vitamin D levels in patients with gestational dia- ous subcutaneous insulin infusion (CSII) over multiple daily injections (MDI) betes mellitus (GDM) and explore the relationships between vitamin D and in the general T1DM population, however, due to limited evidence the supe- pancreatic beta-cell function. riority of CSII over MDI in pre-gestational diabetes (PGDM) has not been Methods: In total, 376 pregnant women were enrolled and divided into unequivocally determined. Therefore, we performed a systematic review GDM group (n = 150) and healthy control group (n = 226), Homeostasis model based on RCTs and non-RCTs to evaluate the effi cacy and safety of CSII assessment 13-cell function (HOMA-B), homeostasis model assessment vs. MDI in women with PGDM. The systematic search retrieved 38 origi- insulin resistance index (HOMA-IR), general clinical characteristics and dia- nal papers, including 15 conference abstracts. The majority of studies were betes-related indexes were tested and compared. The correlations of vita- open label, non-RCTs with high risk of selection bias. Overall, the studies min D with blood glucose, homeostasis model assessment β-cell function reported on 4499 pregnancies complicated by T1DM, out of which 183 were

ADA-Supported Research For author disclosure information, see page A696.

A589 PREGNANCY—CLINICAL/EPIDEMIOLOGYCATEGORY

included in RCTs. Meta-analyses of available studies has shown that the dif- signifi cantly higher (258±58 vs. 230±50 µmol/L, p<0.005) and correlated ference between CSII and MDI with respect to HbA1c was largest in the 1st with BMI and the severity of carbohydrate disorders. Comparison of normal trimester (WMD = -0.45% [-0.64; -0.26)] and became less pronounced over weight and normoglycemic women (79 with pGDM and 24 from C group) also the course of pregnancy, reaching the value of WMD = -0.24% [-0.39; -0.10] demonstrated higher glycemia in OGTT and higher UA levels in pGDM group and WMD = -0.11% [-0.22; -0.001] in the 2nd and the 3rd trimester, respec- (232±48 vs. 208±48 µmol/L, p<0.05). Multivariate analysis revealed signifi - tively. In the CSII group weight gain during pregnancy was higher while daily cant correlations of UA levels with BMI, creatinine, triglycerides and family insulin dose was lower in all 3 trimesters. The difference between CSII and history of diabetes. In logistic regression analysis UA levels ≥ 297 µmol/L MDI was not statistically signifi cant in respect to the risk of severe hypogly- (still within normal range) were associated with the presence of any car- caemia (RR = 0.92 [0.67; 1.26)] or ketoacidosis (RR = 1.58 [0.85; 2.95)]. The bohydrate metabolism disorder (OR 3.62 [95% CI 1.8-7.3], p<0.001). Higher percentage of neonates with birth weight large for gestational age (>90th UA level may be associated with the development of glucose metabolism percentile) was signifi cantly higher in the CSII group (RR = 1.20 [1.07; 1.35)]. abnormalities in pGDM women. As it is observed also in normal weight and In conclusion, the use of CSII during pregnancy complicated by T1DM may normoglycemic women it seems to be a primary disorder. result in improvement of glycemic control as compared to MDI, however, Supported By: Ministry of Science and Higher Education (N 402 069 32/2047) it seems to be associated with larger maternal weight gain and the risk of macrosomia. 2345-PUB Is Real-Time Continuous Glucose Monitoring during Pregnancy in 2343-PUB Women with Type 1 Diabetes Mellitus Related to Better Glycemic Rates and Types of Postpartum Diabetes Testing in Women with Control and Key Obstetric Outcomes? Gestational Diabetes Mellitus: A Population-based Study in Alberta, HAROLD W. DE VALK, JUDITH VAN NIEL-STAAKMAN, GERARD H. VISSER, Canada PETRONELLA N.M. GEELHOED-DUIJVESTIJN, Utrecht, Netherlands, The Hague, SONIA BUTALIA, LOIS DONOVAN, ANAMARIA SAVU, ALUN EDWARDS, JEFFREY Netherlands A. JOHNSON, PADMA KAUL, Calgary, AB, Canada, Edmonton, AB, Canada Real time continuous glucose monitoring (rtCGM) has been shown to There is some evidence to suggest that guideline recommended postpar- improve glycemic control. Limited data are available on the effects of rtCGM tum testing for diabetes in women with gestational diabetes mellitus (GDM) during pregnancy. The current study assesses glycemic control and obstet- is sub-optimal. We assessed the rate and type of postpartum testing in a ric outcomes in women with type 1 diabetes, comparing rtCGM with usual large population level pregnancy cohort in Alberta, Canada., We examined self-measurement of blood glucose (SMBG). Patients were recruited from Therapeutics the extent to which testing was modulated by patient and pregnancy char- two hospitals. All patients were eligible. Glycemic control was assessed by Clinical Diabetes/

PUBLISHED ONLY acteristics. preconceptional HbA1c and HbA1c during pregnancy. Obstetric outcomes Our study consisted of 132,905 birth events, gestational age ≥ 29 weeks, reported are preterm delivery (PD, < 37 weeks), and macrosomia (Large for between 10/01/2008 and 12/31/2011. Laboratory data were used to identify Gestational Age (LGA; birthweight ≥90th percentile; Very Large for Ges- women with impaired glucose tolerance (IGT) or GDM during pregnancy, as tational Age (VLGA; birth weight ≥ 97.7th percentile). Sixty-seven women well as to examine postpartum testing for diabetes within 1 year of deliv- were included; 53 (79%) used rtCGM. Women were completely free in their ery. Multivariable logistic regression was used to identify factors associated choice between rtCGM or SMBG. PD occurred in 22% of the pregnancies, with postpartum testing. 19.6% with RTCGM, 31% on SMBG (p=0.5). LGA occurred in 58% of pregnan- The mean age of women with IGT (n=3,669) and GDM (n=5,034) was 31.7 cies, VLGA in 31%. LGA with RTCGM 61%, with SMBG 46% (p=0.4). VLGA (+/- 5.2 SD) years and 32.0 (+/- 5.3 SD) years, respectively (p<0.01). Over- with RTCGM 37.5%, SMG 50%, p=0.9). Mean HbA1c was signifi cantly lower all, 55.1% of women underwent postpartum diabetes testing. Testing rates before pregnancy (52.1 ± 6.1 vs. 67.8 ± 18.4 mmol/mol, p<0.001) and during were higher among women with GDM (58.3%) than women with IGT (50.7%, the fi rst trimester (6 weeks: 47.9 ± 5.4 vs. 58.9 ± 18.1 mmol/mol, p=0.002 p<0.01). Women were tested using an oral glucose tolerance test (59.7%), a and 12 weeks 43.9 ± 7.0 mmol/mol vs. 53.6 ± 12.8 mmol/mol. p<0.05), but hemoglobin A1c (17.4%), and/or a fasting or random only (22.9%). not later in pregnancy. In this exploratory study, rtCGM was associated with Among women who were tested, 5.5% were diagnosed with diabetes. Fac- better early HbA1c. Preterm delivery may occur less frequently with rtCGM, tors associated with higher odds for postpartum testing were: GDM relative but not macrosomia. Macrosomia may even occur more frequently, possi- to IGT (OR 1.36, 95%, CI 1.25 to 1.49), increasing maternal age (OR 1.03, 95% bly related to better initial glycemic control and better placentation. Future CI 1.02 to 1.04), null-parity (OR 1.23, 95% CI 1.12 to 1.35), presence of a pre- analysis with this expanding group will assess whether these differences pregnancy medical condition (OR 1.18, 95% CI 1.02 to 1.37), induction of labor persist; at this moment, better early control is not readily associated with (OR 1.35, 95% CI 1.22 to 1.48), and urban residence (OR 1.53, 95% CI 1.33 to less macrosomia. 1.76). Smoking during pregnancy was associated with a lower likelihood of postpartum testing (OR 0.55, 95% CI 0.48 to 0.63). 2346-PUB Postpartum testing rates remain sub-optimal and continue to be infl u- enced by maternal and pregnancy characteristics. Strategies to improve WITHDRAWN postpartum testing are warranted in these high-risk women. Supported By: Canadian Institutes of Health Research 2347-PUB 2344-PUB Prevalence of Thyroid Dysfunction and Antithyroid Peroxidase Uric Acid and Glucose Metabolism Disorders in Women with Previ- Antibody in Gestational Diabetes Mellitus ous Gestational Diabetes Mellitus CHANDNI R., ARUN KA, SUNEETHA KALAM, THULASEEDHARAN NK, Kozhikode, PIOTR MOL DA, ANETA FRONCZYK, KRZYSZTOF SAFRANOW, LILIANNA India MAJKOWSKA, Police, Poland, Szczecin, Poland Gestational diabetes mellitus (GDM) and thyroid dysfunction are known to Previous gestational diabetes mellitus (pGDM) is associated with a higher cause adverse maternal and fetal outcome. This study was done to deter- risk of type 2 diabetes (DM2) and higher cardiovascular risk. Uric acid (UA) mine the prevalence of abnormal thyroid function and anti Thyroid Peroxi- is a strong, independent risk factor for both of these diseases. Its role in dase antibody (anti TPO) positivity in patients with GDM and its risk factors. the development of DM2 in women with pGDM remains unclear. The aim A cross sectional study with follow up at 6 weeks post partum was done in of the study was to evaluate the UA levels in pGDM women in relation to 100 consecutive pregnant women diagnosed to have GDM based on 75g oral their current nutritional statuses and . 199 women GTT at 24-28 weeks of gestation. TSH, Free T3, Free T4 and anti TPO anti- with pGDM diagnoses based on oral glucose tolerance tests (OGTTs) 5-12 body were done in all study subjects. The mean age was 28.13 ± 4.80 years. years previously and a control group of 50 women without pGDM (C). The Abnormal thyroid function was detected in 31 (31%) patients (Table). Anti assessment included anthropometric parameters, body composition (Tanita TPO antibody was positive in 35 patients (35%). Increase in age and parity SC-330S), current OGTT, insulin resistance index (HOMA-IR), β-cell function was found to be associated with increased prevalence of anti TPO (p <0.05). (HOMA-%B), HbA1c, lipids, and serum UA. No differences between groups Family history of type 2 DM and GDM (21%), past history of GDM (8%), posi- were found in terms of age, time from indexed pregnancy, anthropometric tive Anti TPO (35%) were signifi cantly associated with thyroid dysfunction parameters, lipids or creatinine levels. Carbohydrate abnormalities were (p <0.005). The presence of thyroid swelling was found to be signifi cantly more frequent in the pGDM group (43.2% vs. 12.0% p<0.001). The women associated with thyroid dysfunction (p <0.005). Nine babies had macrosomia with pGDM had signifi cantly higher fasting glucose, HbA1c, glucose and (>4 Kg) of which 3 mothers had hypothyroidism (p < 0.05). The prevalence insulin levels in the OGTTs, but similar HOMA-IR values. Their UA levels were of thyroid dysfunction and anti TPO antibody was found to be high in GDM

For author disclosure information, see page A696. ADA-Supported Research

A590 PREGNANCY—CLINICAL/EPIDEMIOLOGYCATEGORY patients compared to normal pregnant women. Patients with family history normal range=190-270 µmol/L). Our results suggest a potential relation- of diabetes, past history of GDM, anti TPO positivity, thyroid swelling are at ship of vitamin D insuffi ciency and increased OPN in GDM but whether this high risk for thyroid dysfunction. association enhances the development of insulin resistance during preg- Table. Thyroid Dysfunction and Anti-TPO Positivity. nancy or aggravates insulin resistance in GDM requires further study. In addition, since elevated OPN has been shown to be linked to preeclamp- Number of subjects Anti TPO positive (%) sia, further investigation is needed to determine if increased OPN in the Hypothyroidism 10 90 presence of 25(OH)D insuffi ciency is associated with maternal/neonatal Subclinical Hypothyroidism 17 41 complications. Subclinical Hyperthyroidism 2 0 Isolated Low T3 1 100 2350-PUB Changes in Preconception Treatment and Glycemic Control in Isolated Low T4 1 100 Women with Type 1 Diabetes Mellitus: 15 Years, One Centre Obser- Total 31 58 vation Supported By: Research Society for the Study of Diabetes in India, Kerala Chapter KATARZYNA CYGANEK, JAN SKUPIEN, ALICJA HEBDA-SZYDLO, BARBARA KATRA, IZABELA JANAS, IWONA TRZNADEL-MORAWSKA, PRZEMYSLAW WITEK, ELZBI- ETA KOZEK, MACIEJ T. MALECKI, Krakow, Poland 2348-PUB Background: Pregnancy complicated by type 1 diabetes (T1DM) is asso- A Validation Study of Self-Reported Gestational Diabetes Diagnosis ciated with a high risk of obstetric and neonatal complications. Pregnancy and Treatment planning and excellent glycemic control prior to conception help limit the STEFANIE N. HINKLE, PAUL ALBERT, SHRISTI RAWAL, YEYI ZHU, CUILIN ZHANG, number of outcomes. Rockville, MD Aim: We assessed clinical characteristics of T1DM women, their pre-con- Self-report of gestational diabetes (GDM) is often used in identifying ceptional medical care and glycemic control during fi rst pregnancy visit over GDM events in epidemiological studies; however, data on its validity are a period of 15 years. inconsistent and limited among a multi-racial population. We aimed to Materials and Methods: We analyzed medical records of 524 pregnant assess the validity of self-reported GDM diagnosis 6-weeks postpartum T1DM women, who received diabetes care during pregnancy between 1998 (range 2-12 wk) among a racially and socioeconomically diverse sample and 2012. This period was analyzed as three 5-year intervals. We assessed and to further evaluate the accuracy of self-reported results from the glu- Therapeutics details of patients’ characteristics, use of various therapeutic tools, glycae- cose challenge test (GCT) and GDM treatment. Data were obtained from Clinical Diabetes/

mic control as assessed by the HbA1c level. PUBLISHED ONLY a nested GDM case (n=102)-control (n=101) study within the prospective Results: We did not observe differences in the women’s age (mean 28.4 NICHD Fetal Growth Study. Oral glucose tolerance test results from medi- for the entire study group), T1DM duration (mean 11.8 years), pre-pregnancy cal records were applied as the gold standard for GDM diagnosis. Sensitiv- BMI (mean 24.1 kg/m2), and pregnancy week at the 1st visit (8.7). The number ity was 97.1% [95% confi dence interval (CI) 91.7-99.0] and specifi city 100% of women planning pregnancy did not change signifi cantly and it reached [95% CI 96.3-100] with only three cases reporting not having GDM and no 32.1% in the 1st period (1998-2002), 44.4% in the 2nd period (2003-2007) and controls reporting having had GDM. Sensitivity varied only across mater- 40.3% in last period (2008-2012); (p=0.3). The number of women treated nal race-ethnicity with 100% sensitivity among blacks and Hispanics and with modern technologies increased rapidly: the use of rapid-acting ana- 95.7% and 92.0% among whites and Asians, respectively (P=0.045). Only logues of insulin went up from 2.6% to 46.5% and 95.6%; (p<0.001). Simi- half of the women attempted to recall their GCT values (n=95/203; 46.8%), larly, personal pumps before pregnancy were available only to 4.6% of T1DM but this was not dependent on GDM status. The average self-reported GCT women in the fi rst period, which increased to 23.5% and 33.3% in the 2nd and levels did not differ from those in medical records (median bias: 0.0 mg/ 3rd period, respectively; (p<0.001). We observed a decrease of HbA1c level at dL; interquartile range: 0, 0). The accuracy of reported test results did not the 1st pregnancy visit over the study period from 7.4±1.6% (57mmol/mol) to differ materially by maternal sociodemographic characteristics. According 7.1±1.5% (54mmol/mol) and 7.0±1.4% (53mmol/mol); (p<0.003). to the medical record, most cases of GDM were treated by diet control Conclusions: We observed a rise in the use of modern therapeutic tools (n=63, 63.6%), but a third (n=33) by medication. Almost all (n=62; 98.4%) of (insulin analogs, personal pumps) in T1DM women before their pregnan- the diet controlled GDM women self-reported that they had diet and life- cies and an improvement in glycemic control during the 1st pregnancy visit; style modifi cation and most (n=28; 84.9%) women with medication treated however, the proportion of women planning their pregnancies remained GDM indicated that they had insulin or another medication as the treat- stable. ment. In conclusion, at 6-weeks postpartum, women can accurately recall whether they had GDM and their treatment method if diagnosed. How- ever, most women are unable to recall GCT levels, but were mostly accu- 2351-PUB rate when reported. WITHDRAWN 2349-PUB Vitamin D Insuffi ciency and in Gestational Diabetes Mellitus 2352-PUB RAELENE E. MASER, M. JAMES LENHARD, RYAN T. POHLIG, Newark, DE HbA1c Had a Strong Correlation with the Introduction of Insulin Osteopontin (OPN), a multifunctional molecule, plays a role in fetal growth Therapy during Pregnancy in Japanese Patients with Gestational during pregnancy and may be associated with the vitamin D pathway in fetal Diabetes Mellitus life. In addition, OPN may be causally involved in the pathogenesis of type JIRO KATO, NORIO OKADA, JUNJI SHINODA, Toyota, Japan 2 diabetes (T2D) via a pathophysiological role of infl ammatory processes The current study aimed to identify optimal factors which predict the associated with obesity-induced adipose infl ammation and insulin resis- introduction of insulin therapy during pregnancy in gestational diabetes tance. Given a potential association of OPN and T2D, the fact that gesta- mellitus (GDM) patients. The optimal predictors might contribute to lower tional diabetes mellitus (GDM) is a state of hyperglycemia due to decreased medical expenses and improve quality of their lives during pregnancy. 126 insulin sensitivity during pregnancy, vitamin D may promote beta-cell func- pregnant women were diagnosed with GDM in 75g-OGTT according to the tion and insulin sensitivity, and low levels of 25-hydroxyvitamin D [25(OH)D] IADPSD criteria at our hospital. Of the participants, insulin therapies were are prevalent in GDM, we evaluated whether 25(OH)D insuffi ciency (i.e., initiated in 53 patients whose 2 hour-postprandial plasma glucose levels <30 ng/mL) was independently associated with OPN in women with GDM. were above 7.8mmol/l. HbA1c, fasting plasma glucose, and 1 hour or 2 hour- Individuals (n=31) were 32±4 years old, had a BMI=35±8 kg/m2, and were plasma glucose concentration after 75g oral glucose load, were evaluated at at 30-35 weeks gestation. The prevalence of 25(OH)D insuffi ciency/defi - the time of their diagnosis. Additionally, BMI, maternal age, and gestational ciency was 71%. OPN levels were higher (89±28 vs. 65±24 ng/mL, p<0.05) age were also included to analysis. The statistical analysis of correlations for those that were 25(OH)D insuffi cient/defi cient compared to those that between each of these factors and the introduction of insulin therapies were were not. Linear regression modeling with OPN as the dependent variable performed using Fisher’s exact test and Logistic regression analysis. The revealed that 25(OH)D insuffi ciency (p=0.015) was independently associ- results of these analyses were shown in the Table. HbA1c and 1 hour- or 2 ated, while adjusting for BMI (p=0.042), (model R2=0.26, p=0.014). Although hour-plasma glucose concentration after 75g oral glucose load had a corre- hyperglycemia can upregulate OPN, it should be noted that individuals lation with the introduction of insulin therapy during pregnancy. Especially, in this cohort were in excellent glycemic control (fructosamine=180±15; HbA1c level had a strong correlation In conclusion, higher HbA1c levels at the

ADA-Supported Research For author disclosure information, see page A696.

A591 PREGNANCY—CLINICAL/EPIDEMIOLOGYCATEGORY

time of diagnosis of GDM might predict the needs of insulin therapies during Table. Risk Factors for Postpartum Glucose Intolerance in Women Diagnosed the course of pregnancy. with GDM by Abnormal FBG. Table. Normal group Glucose P-value (n=139) Intolerance group (n=33) Maternal age (years), mean ± SD 32.80 ± 6.04 34.83 ± 5.25 0.060 Prepregnancy BMI (kg/m2), mean ± SD 29.71 ± 6.61 32.12 ± 5.79 0.020 Weight gain during pregnancy (kg), mean ± SD 7.39 ± 6.54 9.25 ± 5.71 0.180 Family history of diabetes, n (%) 85 (61.2) 22 (66.7) 0.557 Chronic Arterial Hipertension, n (%) 32 (23.0) 13 (39.4) 0.054 FBG at diagnosis (mg/dL), mean ± SD 97.72 ± 6.06 101.73 ± 7.40 0.001 2353-PUB Insulin requirement during pregnancy, n (%) 51 (36.7) 23 (69.7) 0.001 Diagnosis of Prediabetes and Diabetes Mellitus in Pregnancy PERCY N. PACORA, Lima, Peru The purposes of this study were: 1.) to defi ne the diagnosis of prediabetes and diabetes mellitus (DM) in pregnancy and 2.) to demonstrate the maternal 2355-PUB Retrospective Study of Progression of Diabetic Retinopathy in Type 1 and perinatal outcome of maternal hyperglycemia in pregnancy. Diabetes We conducted a retrospective study of 5,817 singleton pregnancies that MANDANA MOOSAVI, CLAIRE HARRIS, HEATHER CADENHEAD, ANDREA PAT- underwent a 2h-75g-OGTT during pregnancy at San Bartolome Hospital in ERSON, SHARON E. THOMPSON, DAVID M. THOMPSON, Vancouver, BC, Canada Lima, Peru from January 2000 to December 2012. Maternal glycaemia at Pregnancy and poor glucose control are associated with progression of fasting, 1 h and 2 h after 2h-75g-OGTT were classifi ed as follows: Normo- retinopathy. It is not clear if rapid improvement in glucose control in early glycemia (0 ɖ percentile ɖ 75), prediabetes (75

Therapeutics Information System. IRB approval was obtained. th This was a retrospective study of 176 women with T1DM who delivered at

Clinical Diabetes/ The 75 percentile of maternal glucose level at fasting, 1 h and 2 h corre- BC Women’s hospital between December 1999 and March 2015. Eligible PUBLISHED ONLY sponded to 84, 137 and 108 mg/dl, respectively. The 95th percentile of mater- patients had an HbA1c measurement and ophthalmology assessment either nal glucose level at fasting, 1 h and 2 h corresponded to 97, 175 and 137 mg/ immediately before pregnancy or in the fi rst trimester at 4 ± 7 weeks (mean ± dl, respectively. standard deviation) gestational age (GA) and repeated at 25 ± 6 weeks. Retin- The prevalence of DM, prediabetes and normoglycemia in the study group opathy was assessed by 2 ophthalmologists and progression defi ned as an were 9.3%, 35.9% and 54.8%, respectively. Women with prediabetes had increase by 1 or more steps on the International Clinical Diabetic Retinopa- 2 folds increased risk of being diagnosed as DM during pregnancy (IC 95%: thy Scale. The patients were on average age 26 ± 1.3 ± with 11± 3 years with 1.67-2.39). As compared with normoglycemic women, prediabetes and DM T1DM. The mean initial HbA1c was 7.2% ± 2.5 (range 4.2 to 12.5%). Out of patients were older, heavier, had more frequency of being married, being 173 patients, 55 had progression of DR and 118 did not. To examine if base- multiparous, had poor obstetric history, history of vascular disease, pri- line glucose control affected the risk of progression, we defi ned poor con- mary cesarean section, neonatal birth weight >4 kg, preterm neonate, large trol as entry A1c > 8.0% and good control as ɖ 8.0%. Progression occurred for gestational age-infant (p< 0.01) and increased rate of neonatal sepsis in 57.5% of patients in poor control compared with 28.5% with good con- (p<0.05). trol and this difference did not reach statistical signifi cance (p = 0.11). The Hyperglycemia in pregnancy may be diagnosed using one glucose value severity of baseline retinopathy also did not correlate with risk of progres- greater than the 75th percentile in a 2h-75g- OGTT. These maternal glycemia sion. Progression of retinopathy occurred in 25% with mild or no DR com- levels at fasting, 1 h and 2 h correspond to 84, 137 and 108 mg/dl, respec- pared with 30% in those with moderate or greater DR (p=0.217). We then tively. examined if rapid improvement in blood glucose control, defi ned as a drop Prediabetes should be diagnosed in pregnancy because, like DM, it is in A1c of >1% from baseline, affected the risk of progression. Progression associated with signifi cant increased risk of primary cesarean section and occurred in 52.7% of those who had a rapid drop in A1c compared with 25% perinatal morbidity. who did not quickly lower their A1c and this difference was highly signifi cant (p = .0005). We conclude that a rapid lowering of A1c in early pregnancy is 2354-PUB associated with an increased risk of progression of diabetic retinopathy. Risk Factors for Postpartum Glucose Intolerance in Patients with GDM Diagnosed by Abnormal Fasting Blood Glucose 2356-PUB ANA CRLA SOUZA, RAFAELA A. COSTA, CRISTIANE F. PAGANOTI, MARCELO The Roles of Inappropriate Gonadotropin Secretion and Insulin ZUGAIB, ROSSANA P.V. FRANCISCO, São Paulo, Brazil Resistance in the Pathogenesis of Hyperandrogenism in Polycystic Gestational diabetes (GDM) is a major risk factor for diabetes mellitus Ovary Syndrome during a woman lifetime and has increased among pregnancies, especially DONG XUEJIE, LIU JIA, HU YANJIN, XU YUAN, WANG GUANG, Beijing, China after diagnostic criteria included women with abnormal fasting blood glu- Polycystic ovary syndrome (PCOS) is an endocrine disease in reproductive- cose (FBG) at the fi rst prenatal visit (IADPSG consensus). We studied risk fac- aged women. Hyperandrogenism (HA) is an important feature of PCOS. The tors for postpartum glucose intolerance persistence in GDM women diag- main cause of PCOS are inappropriate luteinizing hormone (LH), follicle stim- nosed by abnormal FBG at fi rst prenatal visit by performing a cross-sectional ulating hormone (FSH) and LH/FSH ratio (>2). Another epiology of PCOS is study that included subjects who attended prenatal care in a tertiary teach- insulin resistance (IR) with followed hyperinsulinemia. Both of them can lead ing hospital (Sao Paulo, Brazil) from 2012 to 2014. We reviewed the medi- to hyperandrogenism and PCOS. Our research tried to detect the association cal records from GDM women with singleton pregnancies, who presented of IR, inappropriate gonadotropin and hyperandrogenism in PCOS patients. abnormal FBG ( 92mg/dL) before 24 gestational weeks and who performed ≥ 270 newly diagnosed PCOS patients were recruited in accordance with the postpartum oral glucose tolerance test (OGTTp) with 75g of glucose 6 the guidelines of Rotterdam ESHRE/ASRM Consensus 2003. Patients were to 12 weeks after parturition. Primary outcome was the persistence of any divided in 2 groups: 215 with LH/FSH 2 and 55 with LH/FSH<2. Patients degree of glucose intolerance at OGTTp (FBG 100mg/dL and/or 2 h after ≥ ≥ were interviewed to obtain history records along with the data of physical overload 140mg/dL). A total of 172 women were studied and 33 (19, 18%) ≥ and laboratory examinations. A 75g oral glucose tolerance test was per- remained with postpartum glucose intolerance. Clinical and laboratorial formed. IR was evaluated using homeostasis model (HOMA-IR). The corre- data were compared between groups, according to maintenance of glucose lations of androstenedione (A) and total testosterone (T) and fasting insu- intolerance. Higher prepregnancy BMI and higher FBG levels at diagnosis lin, HOMA-IR and LH/FSH were included. Our results showed that the body were risk factors related to that outcome. Insulin requirement during preg- mass index, triglycerides, glucose and fasting insulin were signifi cantly nancy increased by almost 4x the odds of postpartum glucose intolerance higher in patients with LH/FSH<2. HOMA-IR [1.71 (0.72, 3.23) vs. 0.98 (0.44, (OR 3.97, 95% CI 1.75-9.00). 2.15), P<0.05] was signifi cantly higher in LH/FSH<2 group. In the LH/FSH≥2 group, A and T were positive correlated with LH/FSH (A: r=0.272, P=0.049;

For author disclosure information, see page A696. ADA-Supported Research

A592 PREGNANCY—CLINICAL/EPIDEMIOLOGYCATEGORY

T: r=0.297, P=0.000). In the LH/FSH<2 patients, A and T were positive cor- overall, PE was associated with increased BW (p=0.015) and a non-signifi - related with LH/FSH (A: r=0.297, P=0.000; T: r=0.252, P=0.000), HOMA-IR cant reduction in PWR in DM. In diabetes, high fetal weight and lower PWR (A: r=0.290, P=0.000; T: r=0.289, P=0.000) and fasting insulin (A: r=0.256, may provide “fertile soil” for PE. P=0.000; T: r=0.272, P=0.000). We concluded that IR was more serious in Figure. PCOS with LH/FSH <2. In the PCOS patients, HA was correlated with the LH/FSH ratio. And in the LH/FSH<2 group, HA was also correlated with IR and hyperinsulinemia. This group of PCOS patients may have higher risk of metabolic diseases. Supported By: National Natural Science Foundation of China (81270369, 81070244, 30770873)

2357-PUB WITHDRAWN

2358-PUB Patient Experiences with and Opinions on Real-Time Continuous Glucose Monitoring in Pregnancies Complicated by Type 1 Diabe- tes Mellitus HAROLD W. DE VALK, MEREL S. STOLZE, LETTE BEA HOEKS, BIANCA SILVIUS, Supported By: National Institute on Minority Health and Health Disparities GERARD H.A. VISSER, Utrecht, Netherlands Real time continuous glucose monitoring (rtCGM) improves glycemic con- 2360-PUB trol in non-pregnant patients. Less is known about rtCGM in pregnant dia- Hypertensive Disorders of Pregnancy in Women with a History of betic women. Analysis of personal experiences provides information to opti- Gestational Diabetes Signals Incident Hypertension in Partners

mally implement rtCGM during pregnancy. ROMINA PACE, ELHAM RAHME, KABERI DASGUPTA, Montreal, QC, Canada Therapeutics

Women delivering in 2012 using rtCGM were interviewed using a struc- Like hypertension and type 2 diabetes, hypertensive disorders of preg- Clinical Diabetes/ tured questionnaire on user preferences, technical problems, effects on nancy (HDP) and gestational diabetes (GDM) are linked to eating patterns, PUBLISHED ONLY daily life and care received. rtCGM is reimbursed during pregnancy in the physical activity levels, socioeconomic status, and other behavioral and Netherlands and women are complete free in their choice between rtCGM environmental factors. These are shared between spouses. We previously and usual care without rtCGM. demonstrated (1) spousal concordance for diabetes and (2) GDM to oper- Twenty women with type 1 diabetes delivered in 2012; 12 (60%) chose ate as a risk factor for incident diabetes in male partners. In the present rtCGM, all on CSII, 10 (83%) started during the fi rst trimester. Most (11 analyses, we evaluated the impact of HDP in women on incident hyperten- women, 92%) pregnancies were planned. Education: university in 4, higher sion in their partners. We separately evaluated couples with and without a vocational in 5, middle vocational in 3. Two-thirds used rtCGM continuously; GDM history. We analyzed data from 67,394 Quebec couples (health admin- others allowed themselves a few days off. Two-thirds needed 1-4 weeks to istrative and birth registry data). Half had a GDM history, between April 1st, learn to use rtCGM; blood glucose was measured on average 3-5 times a day. 1990 and December 31st, 2007, and were matched (age group, health region, Social and working activities did generally not suffer from RT-CGM. Techni- year of delivery) with couples without such a history. Follow-up data was to cal problems were minimal; delay between changes in plasma and intersti- March 31st, 2012 for both fathers and mothers. In the present analyses, we tial glucose values were frequent mentioned practical problem, especially excluded couples with a hypertension diagnosis in the prior 3 years. Overall, in the lower range. Most women continued rtCGM during lactation and all 11.7% of fathers had a partner diagnosed with HDP. The fathers whose part- would use rtCGM in a next pregnancy. All women expressed that care was ners had HDP were older with a greater number of prior pregnancies. HDP in best given by a small team well-attuned to their condition. partners conferred a 16% risk increase in fathers whose partners had GDM rtCGM in pregnancy in well-educated women with type 1diabetes is (Odds ratio, OR, 1.16; 95% CI 1.07, 1.26) but did not appear to increase risk in associated with high maternal compliance, minimal effect on daily life and fathers whose partners did not have GDM (OR 1.02, 95% CI 0.91, 1.15). OR greatly appreciated with about half of the pregnant women wanting to use was adjusted for age, age of partner, previous pregnancy with partner, living it. Delay in timely detecting hypoglycemia is a major issue and optimal use with partner at delivery, hospitalization in prior 3 years, deprivation index of rtCGM requires care by a small, local team. These results can help to opti- and ethnocultural background. mize clinical care and may also help to design patient education and guid- ance in clinical trials. 2361-PUB Evaluation of Physical Activity Performed between the Main Meals 2359-PUB for Improving Glucose Control in Women with Gestational Diabetes Effects of Diabetes and Preeclampsia on Birth and Placenta Weight Mellitus (GDM) in American Indian and Hispanic Pregnancies PIERGIORGIO FRANCIA, ELENA BIANCHI, MASSIMO GULISANO, ANNA TEDE- MISTI J. LEYVA, CHRISTOPHER E. ASTON, LANCER D. STEPHENS, MADONA G. SCHI, ROBERTO ANICHINI, ALESSANDRA DE BELLIS, Florence, Italy, Pistoia, Italy AZAR, RICHARD MCCLAIN, KEITH GOODMAN, TIMOTHY J. LYONS, Oklahoma The correct management of physical activity (PA) can improve glucose con- City, OK, Ada, OK, Talihina, OK, Belfast, United Kingdom trol in patients with GDM. The aim of this study is to evaluate the usefulness Preeclampsia (PE), characterized by de novo hypertension and proteinuria of controlling the effect of PA performed between the 3 main meals in terms in pregnancy, is a major cause of premature delivery, and maternal and fetal of glucose control and nocturnal sleep quality (NSQ). In 8 patients with GDM, death worldwide. PE is increased in pregnancies complicated by dysglyce- mean age 36.1±4.1 yrs, week of gestation 26.3±7.3, in therapy with diet and mia, especially pre-gestational diabetes (incidence ~20% vs. ~5% in general insulin (4) and with diet (4), and in 8 age-, sex-, matched healthy controls we population). The placenta is central to PE development. Recent reports indi- evaluated: the daily PA divided into 3 different periods of the day: Evening (9 cate that altered placental weight and/or placental weight ratio (PWR: pla- p.m. to 7.59 a.m.), Morning (8 a.m. to 1.59 p.m.) and Afternoon (2 p.m. to 8.59 cental weight (PW)/birth weight (BW): refl ects balance between fetal and p.m.), and NSQ for 1 week […] by Actigraph GT9X Link, sleep and PA diary. placental growth) may predict obstetric outcome. Comparing PW, BW, and Patients with GDM received diet and glucometer for monitoring glycemia PWR between pre-gestational type 2 diabetic (DM), gestational diabetic while fasting and 1 hour after meals. The patients’ daily PA and its distribu- (GDM), and normoglycemic (NG) pregnancies, we report data from samples tion in the 3 different day periods (E.M.A.) was not signifi cantly different collected in an ongoing, prospective study of dysglycemia and PE in Ameri- from that of controls. Patients had poorer NSQ (76.52%±11.6) than controls can Indian and Hispanic women, two groups at increased risk for DM. Of 145 (92.8%±2.4;p<0.005), which was correlated to the PA (steps number) per- available placentas (50 DM, 68 GDM, 27 NG), 9 were from PE pregnancies formed in the previous A period (r=0.86, p<0.01) and inversely correlated (4 DM, 3 GDM, 2 NG). PW, BW and PWR were adjusted for gestational age. to the sedentary lifestyle in the same length of time (r=0.89, p<0.01). Total BW was signifi cantly higher among normotensive women with DM vs. GDM evaluated glycemia was correlated with the amount and intensity of PA per- (p=0.005) or NG (p=0.007); PW was similar in all three groups; PWR was formed in the previous E.M.A. period (steps number: r=0.29, p<0.01; light: slightly reduced in DM women. The number of PE cases was very small, but r=0.54, p<0.001; moderate: r=0.22;p<0.05), and inversely correlated to sed-

ADA-Supported Research For author disclosure information, see page A696.

A593 PREGNANCY—CLINICAL/EPIDEMIOLOGYCATEGORY

entary time (r=-0.46;p<0.001). Fasting glycemia was inversely correlated to Table. Area Under the Receiver Operating Characteristics (95% CI) to Predict PA (steps number) performed in the previous E period (r=0.45;p<0.05); post- Perinatal Morbidity. prandial glycemia was correlated with moderate PA performed in the previ- Primary Cesarean Birthweight LGA-Infant Neonatal Neonatal Neonatal ous M and A periods (lunch:r=0.52, p<0.01-dinner:r=0.39, p<0.05). The pre- section > 4 kg prematurity sepsis sepsis liminary data of this study suggest that analysis of PA performed between Prevalence 37.67% 11.5% 22.8% 10.8% 3.5% 2.6% the E.M.A. provides useful information on the condition of women with GDM. The E.M.A. method permits better analysis of the PA effect on glu- San Bartolome Criteria 0.536 0.552 0,529 0.562 0.571 0.600 cose control and quality of sleep. (0.513-0.559) (0.528-0.576) (0.511-0.567 (0.545-0.594) (0.508-0.593) (0.551-0.649) WHO 2013 Criteria 0.524 0.527 0.529 0.562 0.551 0.598 2362-PUB (0.501-0.547) (0.503-0.551) (0.511-0.547) (0.537-0.588) (0.508-0.593) (0.547-0.649) Frequency of Gestational Diabetes Mellitus (GDM) in a High-Risk O´Sullivan and Mahan 0.518 0.524 0.529 0.550 0.548 0.590 Obstetric Service Criteria (0.495-0.540) (0.500-0.548) (0.511-0.548) (0.525-0.575) (0.505-0.590) (0.539-0.641) FABIOLA ROMERO, LILIAN ZORRILLA, ALICIA GOMEZ, CARMEN ECHAGUE, ALBA WHO 1980 Criteria 0.515 0.518 0.518 0.542 0.530 0.571 GUERRERO, DIEGO VERA, ZUMILDA IBARRA, NIDIA VILLALBA, ALEJANDRA (0.492-0.538) (0.494-0.542) (0.500-0.536) (0.516-0.567) (0.488-0.572) (0.521-0.622) OCAMPO, HELEN LOPEZ, ELIZABETH VALINOTTI, NADIA GARCIA, MARIA LIS Carpenter and Coustan 0.508 0.514 0.517 0.531 0.539 0.567 ALARCON, LILIANA VIGO, FRANCISCO CABRERA, SANTIAGO BAEZ, ROSA VILLA- Criteria (0.485-0.530) (0.491-0.538) (0.499-0.535) (0.507-0.556) (0.497-0.582) (0.517-0.618) LBA, CAROLINA ACOSTA, MARIA CELIA MENONI, Asunción, Paraguay NDDG Criteria 0.506 0.513 0.513 0.520 0.535 0.553 The prevalence of GDM varies according to different populations, changes (0.484-0.529) (0.489-0.536) (0.496-0.536) (0.495-0.544) (0.492-0.577) (0.502-0.603) in risk factors and the diagnostic criteria used. Currently, ADA recommends the use of the criteria arising from the Hyperglycemia and Adverse Preg- nancy Outcome study (HAPO), which was designed to determine the effects 2364-PUB of hyperglycemia on perinatal morbidity and mortality. There are few data Association between Adolescent Preconception Counseling (PC) on the frequency of GDM according to these new criteria in the Latino and Future Pregnancy Planning and Outcomes as an Adult in Type 1 population.We perform a prospective, longitudinal study of 330 consecu- Diabetes (T1D) tive, inpatients cases from High Risk Obstetrics Service of Hospital Central SUSAN M. SEREIKA, DENISE C. CHARRON-PROCHOWNIK, DOROTHY J. BECKER, Institute Prevision Social, Asunción, Paraguay, from January to July 2013, to

Therapeutics WILLIAM H. HERMAN, PATRICIA L. SCHMITT, ANA DIAZ, ANDREA F. FISCHL, determine the frequency of GDM using ADA criteria and assess the risk fac-

Clinical Diabetes/ Pittsburgh, PA, Ann Arbor, MI

PUBLISHED ONLY tors and the results of the blood sugar curve. There were 160 GDM cases, Pregnant women with diabetes are at increased risk for maternal and corresponding to a frequency of 48%. Age 31 ± 5 years and BMI 26 ± 4 kg/ fetal complications. Planning a pregnancy through PC decreases the risks m2. Among the risk factors: age greater than 30 years 58%, overweight or associated with these complications. The purpose of this study was to com- obese: 50%, family history of diabetes: 35%, previous abortion: 23%, per- pare pregnancy planning and outcomes for the fi rst pregnancy in women sonal history of gestational diabetes or macrosomia: 18%, smoking and alco- who as adolescents had received READY-Girls, an awareness PC program hol consumption: 3%. The OGTT 75 grams glucose was performed at 30 ± 2 for teens with T1D, compared to a matched group of women with T1D who weeks gestation, basal blood glucose values, one and two hours post-load: did not receive READY-Girls as teens. Data were collected via an online self- 96.9 mg/dl, 168.9 mg/dl and 165.4 mg/dl respectively. Only one glucose value reported retrospective and prospective surveys, including age at fi rst receiv- altered in 58% and the distribution was 47%, 9% and 44% for basal glyce- ing PC, pregnancy planning behaviors and pregnancy outcomes. mia, fi rst hour and second hour respectively. Two and three values altered in Results: Of the 102 women followed (18-38 years of age, 97.1% white), 17 21% and 22% respectively. In conclusion, we observed a high frequency of (16.7%) women (8 [47.1%] adolescent PC vs. 9 [52.9%] matched control; 100% GDM, probably because it was a high-risk obstetric population. Among the white) had their fi rst pregnancy prior to or during the study. Women who risk factors maternal age greater than 30 years and obesity were the most received adolescent PC were similar demographically to matched controls common. Regarding glycemia curve, most patients presented a single glu- (mean age at entry [years]: 27.8 vs. 28.3; mean duration of T1D [years]: 18.8 cose value altered, corresponding to the basal glucose, followed by second vs. 20.7; mean age at fi rst pregnancy [years]: 26.0 vs. 24.1; college graduate hour glucose value. However a minimum amount of patients who had iso- or greater: 62.5% vs. 33.3%). Though largely not signifi cant, women receiv- lated altered values of the fi rst hour glucose could be lost by not performing ing adolescent PC tended to be more likely to receive PC as adults (50% the OGTT at 3 points. vs. 12.5%), had less unplanned pregnancy (37.5% vs. 66.7%) and had better pregnancy outcomes (deliveries > 20 weeks gestation: 87.7% vs. 62.5%) 2363-PUB compared to matched controls. Comparing the groups, maternal complica- Effectiveness of Criteria of Diabetes Mellitus in Pregnancy Based tions included A1c > 8 during pregnancy (12.5% vs. 33.3%), preeclampsia on Perinatal Outcome (20.0% vs. 11.1%), and keto-acidosis (0% vs. 11.1%), while infant complica- PERCY N. PACORA, Lima, Peru tions included birth weight > 9 pounds (16.7% vs. 20%), jaundice (16.7% vs. The purpose of this study was to determine the effectiveness for predict- 60%), and severe hypoglycemia (0% vs. 80%, p=.015). No major congenital ing perinatal morbidity of six criteria for diagnosis of diabetes mellitus (DM) anomalies were reported in either group. in pregnancy by oral glucose tolerance test (OGTT). A retrospective study of Conclusion: While the sample is small, most differences between the ado- 5,817 singleton pregnancies was conducted. All pregnancies underwent a lescent PC and matched control groups were as hypothesized. PC started in 2h-75g-OGTT from January 2000 to December, 2012 at Hospital San Barto- adolescence appears to have some benefi cial effects. lomé in Lima, Peru. Maternal and Perinatal outcome were obtained from the Perinatal Information system. IRB approval was obtained. From the study 2365-PUB group, a subgroup of 650 singleton pregnancies of women between 18 to 24 y.o., with BMI <25 kg/m2, nulliparous, without personal/family cardiovas- WITHDRAWN cular risk were selected to obtained the 95th percentile for maternal fast- ing glucose (FG), 1h-glycemia (1h-G) and 2h- glycemia (2h-G) levels of OGTT. These glucose levels were named San Bartolome (SB) Criteria. Effectiveness was calculated by areas under ROC for predicting primary cesarean section, 2366-PUB preterm birth, neonatal birthweight > 4 kg, LGA-infant, neonatal hypoglyce- WITHDRAWN mia and neonatal sepsis using O’Sullivan and Mahan (OM), National Diabe- tes Data Group (NDDG), WHO 1980, Carpenter and Coustan (CC), IADPSG/ ADA 2012 (WHO 2013) and SB Criteria for OGTT. SB criteria consisted on at least one glucose value equal or greater than the following cut-off: FG 93 mg/dl at 1-36 week and 88 mg/dl at 37-41 week; 1h-G 153 mg/dl at 1-32 week, 173 mg/dl at 33-36 week and 150 mg/dl at 37-41 week; 2h-G 121 mg/dl at 1-41 week. SB and WHO 2013 criteria for diagnosis of DM in preg- nancy had better effectiveness for predicting signifi cant perinatal morbidity than the other diagnostic criteria.

For author disclosure information, see page A696. ADA-Supported Research

A594