Combination Therapy with Exenatide Plus Pioglitazone Versus Basal

Diabetes Care Volume 40, March 2017 325

Muhammad Abdul-Ghani,1,2 Combination Therapy With Osama Migahid,1 Ayman Megahed,1 CARE/EDUCATION/NUTRITION/PSYCHOSOCIAL CLIN John Adams,2 Curtis Triplitt,2 Exenatide Plus Pioglitazone Versus Ralph A. DeFronzo,2 Mahmoud Zirie,1 Basal/Bolus Insulin in Patients and Amin Jayyousi1 With Poorly Controlled Type 2 Diabetes on Sulfonylurea Plus Metformin: The Qatar Study Diabetes Care 2017;40:325–331 | DOI: 10.2337/dc16-1738

OBJECTIVE The Qatar Study was designed to examine the efﬁcacy of combination therapy with exenatide plus pioglitazone versus basal/bolus insulin in patients with long-standing poorly controlled type 2 diabetes mellitus (T2DM) on metformin plus a sulfonylurea.

RESEARCH DESIGN AND METHODS

The study randomized 231 patients with poorly controlled (HbA1c >7.5%, 58 mmol/mol) T2DM on a sulfonylurea plus metformin to receive 1) pioglitazone plus weekly exenatide (combination therapy) or 2) basal plus prandial insulin (insulin therapy) to maintain HbA1c <7.0% (53 mmol/mol).

RESULTS After a mean follow-up of 12 months, combination therapy caused a robust decrease 1Diabetes Research, Academic Health System, in HbA from 10.0 6 0.6% (86 6 5.2 mmol/mol) at baseline to 6.1 6 0.1% (43 6 Hamad General Hospital, Doha, Qatar 1c 2 0.7 mmol/mol) compared with 7.1 6 0.1% (54 6 0.8 mmol/mol) in subjects receiving Division of Diabetes, University of Texas Health Science Center at San Antonio, San Antonio, TX insulin therapy. Combination therapy was effective in lowering the HbA indepen- 1c Corresponding author: Muhammad Abdul-Ghani, dent of sex, ethnicity, BMI, or baseline HbA1c. Subjects in the insulin therapy group [email protected]. ﬁ experienced signi cantly greater weight gain and a threefold higher rate of hypogly- Received 11 August 2016 and accepted 19 cemia than patients in the combination therapy group. December 2016. Clinical trial reg. no. NCT02887625, clinicaltrials CONCLUSIONS .gov. Combination exenatide/pioglitazone therapy is a very effective and safe therapeutic This article contains Supplementary Data online option in patients with long-standing poorly controlled T2DM on metformin plus a at http://care.diabetesjournals.org/lookup/ sulfonylurea. suppl/doi:10.2337/dc16-1738/-/DC1. M.A.-G. and O.M. contributed equally to the study. The diagnosis of type 2 diabetes mellitus (T2DM) is made based on the plasma glucose This article is featured in a podcast available at concentration. However, progressive b-cell failure and insulin resistance are the prin- http://www.diabetesjournals.org/content/ cipal core defects responsible for development and progression of hyperglycemia in diabetes-core-update-podcasts. individuals with T2DM (1). Nonetheless, hyperglycemia is the major factor responsible © 2017 by the American Diabetes Association. Readers may use this article as long as the work for retinopathy and nephropathy, and every 1% decrease in HbA1c is associated with ; – is properly cited, the use is educational and not an 35% reduction in risk of microvascular complications (2 5). Microvascular com- for proﬁt, and the work is not altered. More infor- plications are uncommon if the HbA1c is maintained at less than 6.5–7.0% (48– mation is available at http://www.diabetesjournals 53 mmol/mol) (4). Therefore, the standard of diabetes care recommends maintaining .org/content/license. 326 Exenatide Plus Pioglitazone vs. Insulin in T2DM Diabetes Care Volume 40, March 2017

the HbA1c at ,7.0% (53 mmol/mol) to min- than the recommended sequential add- years) treated with a maximal/nearly max- imize the microvascular risk in patients on of metformin, followed by a sulfo- imal dose of metformin (.1,500 mg/day) with T2DM (6). nylurea and insulin (22). Subjects who plus a sulfonylurea (.4 mg glimepiride Metformin is the most commonly used were started on metformin/pioglitazone/ or .60 mg gliclazide) were recruited. first-line antidiabetic agent in T2DM (6). It exenatide at the time of diagnosis of Patients were in good general health lowers HbA1c by inhibiting hepatic glucose T2DM maintained a mean HbA1c of 5.9% as determined by medical history and production (7.8). Because metformin does (41 mmol/mol) for 2 years compared with physician examination. Participants not halt or prevent the progressive b-cell HbA1c of 6.5% (48 mmol/mol) in patients had normal kidney and liver func- failure (9,10), after an initial decrease, the in whom metformin, a sulfonylurea, and tion, serum chemistry results, elec- HbA1c rises progressively over time. Al- insulin were sequentially added (22). The trocardiogram results, and urinalysis, though the American Diabetes Association group with triple therapy with metformin/ stable body weight (6 3 pounds within (ADA) recommends several therapeutic pioglitazone/exenatide also experienced preceding year), and a negative preg- options in patients in whom metformin is much less hypoglycemia and weight gain nancy test. Exclusion criteria were a he- failing, sulfonylureas remains the most than the group receiving sequential addi- matocrit ,34%, medications known to commonly used second-line therapy, and tion of metformin, followed by glipizide affect glucose metabolism other than similar to metformin, they also lack pro- and then a progressively increasing daily sulfonylureas and metformin, evidence tective effect on b-cell function (3,9,10). dose of glargine insulin (22). of diabetic proliferative retinopathy, al- Insulin is very effective in lowering the Millions of patients with poorly con- bumin excretion .300 mg/day, and ma- plasma glucose concentration, but re- trolled T2DM worldwide are currently jor organ system disease, as determined quires an injection and home glucose treated with metformin plus a sulfonyl- by physical examination, medical history, monitoring, causes hypoglycemia, and urea. Although a third oral agent (e.g., and screening blood tests. promotes weight gain (11). Nonetheless, DPP-4 inhibitor) can be added to improve insulin is the recommended therapeutic their glycemic control, because of ad- Study Design option in patients with poorly controlled vanced b-cell failure (1), insulin therapy Eligible subjects were consecutively ran- T2DM when multiple oral agents fail (e.g., is the most commonly used addition in domized by age, sex, BMI, diabetes dura- metformin plus a sulfonylurea) (6). those patients with poorly controlled tion, and HbA1c to receive pioglitazone/ Progressive b-cell failure, weight gain, T2DM on multiple oral agents. exenatide (combination therapy) or basal andhypoglycemiaremainmajorobstacles The aim of the current study was to (glargine)/bolus (aspart) insulin (insu- in maintaining HbA1c within the normal compare the efficacy and safety of com- lin therapy) to maintain HbA1c ,7% range (2,11), including with insulin ther- bination therapy with GLP-1 RA plus thia- (53 mmol/mol). There was no limit on apy. Further, recent clinical trials have re- zolidinedione, which correct the major the upper value of HbA1c for enrollment ported that hypoglycemia is associated metabolic defects responsible for the de- (range of initial HbA1c was 7.5215.1% with increased mortality risk (11). This velopment of progressive hyperglycemia [58–142 mmol/mol]). presents a dilemma in patients with poorly in T2DM (1), versus basal/bolus insulin in Subjects randomized to combination controlled T2DM between lowering HbA1c patients with long-standing poorly con- therapy were started on weekly exenatide to prevent microvascular complications trolled T2DM treated with maximum injection (2 mg/week extended-release and minimizing the risk of hypoglycemia. doses of metformin plus sulfonylurea. exenatide [BYDUREON] administered The ADA recommends “lowering HbA1c with a pen), and pioglitazone (15 mg/day) to ,7.0% (53 mmol/mol) in most patients RESEARCH DESIGN AND METHODS was added at week 2, and the dose was and more stringent targets (6.0–6.5%) The Qatar Study is an open label, single-center, increased to 30 mg/day at week 4. Sub- (42–48mmol/mol)inselectedpatientsif randomized control trial (ClinicalTrials.gov jects receiving insulin therapy were started achieved without hypoglycemia” (6). identifier NCT02887625) that examines on insulin glargine before breakfast. The Thiazolidinediones are potent insulin sen- the efficacy, durability, and safety of Treat-to-Target Trial (4T) algorithm was sitizers (12), improve b-cell function (13–17), combination therapy with exenatide used to calculate the starting glargine and as reviewed by DeFronzo (1), clinical plus pioglitazone versus basal/bolus insu- dose, and the dose was adjusted weekly trials have demonstrated durable HbA1c lin in patients with poorly controlled to achieve a fasting plasma glucose (FPG) reduction with lower risk of hypoglycemia T2DM on metformin plus a sulfonylurea. of ,110 mg/dL. After the FPG goal . compared with sulfonylureas (10,18,19). This is an ongoing study at Hamad General was achieved, if the HbA1c was 7.0% GLP-1 (glucagon-like peptide 1) receptor Hospital, Doha, Qatar. Here we report the (53 mmol/mol), 4–6unitsofinsulinaspart agonists (RA) improve b-cell function (20), effect of each therapy on glycemic control was started before each meal, and the an effect that lasts for up to 3 years (21). and safety parameters at a mean of 1 year dose was adjusted to a achieve a plasma Thus, they produce a durable reduction in of follow-up (range, 6–18 months). The glucose concentration of ,140 mg/dL at HbA1c with a low risk of hypoglycemia. study protocol was approved by the 2hoursaftermeals. We recently demonstrated that the ad- Hamad General Hospital Institutional Re- Patients were seen monthly during the dition of exenatide plus pioglitazone, view Board, and informed written consent first 4 months or as needed, based on the which improve insulin sensitivity and was obtained before patient enrollment. results of the plasma glucose concentra- b-cell function, to metformin in patients tion, and bimonthly thereafter. FPG, body with newly diagnosed T2DM produces a Subjects weight, and HbA1c were measured at greater and more durable reduction in the Subjects with poorly controlled (HbA1c each follow-up visit, and the medica- HbA1c and a lower risk of hypoglycemia .7.5% [58 mmol/mol]) T2DM (age 18–75 tion dose was adjusted to maintain a care.diabetesjournals.org Abdul-Ghani and Associates 327

FPG ,110 mg/dL, 2-h postprandial hypoglycemia was calculated as number 129 were randomized to combination plasma glucose ,140 mg/dL, and HbA1c of subjects experiencing at least a single therapy and 122 to insulin therapy. Six pa- ,7% (53 mmol/mol), unless hypoglycemia event divided by number of patients in tients in the combination therapy arm and (blood glucose was ,60 mg/dL or symp- that arm. 14 patients in the insulin arm dropped out toms) was encountered. Hypoglycemia Values are presented as mean 6 SEM. of the study. This report presents the re- was defined as blood glucose concentration The two-sided t test was used to compare sults of a mean follow-up of 1 year (range, of ,60 mg/dL, with or without symptoms, mean differences between the two treat- 6–18 months) for 123 patients in the com- or hypoglycemia symptoms that subsided ment arms. The x2 testwasusedtotest bination therapy group and 108 in the in- after glucose ingestion. Patients were in- the significance of discrete variables. sulin therapy group. Supplementary Figure structed to perform a 7-point home blood The study was powered to detect a 1 presents the randomization scheme. glucose profile level (One Touch, VivaCheck, 0.55% HbA1c difference between treat- Table 1 presents the baseline character- Wilmington, DE) before and 2 h after a ment arms based on HbA1c difference in istics of study participants. Patients in both meal and at bedtime 1 day each week. our previous Efficacy and Durability of groups were well matched in age, sex, BMI, Blood glucose levels measured at home Initial Combination Therapy for Type 2 disease duration, and baseline HbA1c.Pa- were downloaded at each follow-up visit Diabetes (EDICT) study (22). The EDICT tients were a mean age of 52 6 1, were and verified by the study coordinator. study randomized, patients with new- mildly obese, had poor glycemic control, During each follow-up visit, FPG was mea- onset diabetes to receive initial combina- with mean baseline HbA1c of 10.0% sured, records of glucose values mea- tion therapy with metformin/pioglitazone/ (86 mmol/mol), and mean diabetes du- sured at home were reviewed, and exenatide versus sequential add-on of ration of 10.7 years. patients were questioned about symp- metformin, followed by a sulfonylurea All patients were receiving therapy with toms of hypoglycemia. Severe hypogly- and insulin. At 2 years, .80% of subjects metformin. The mean metformin dose cemia was defined as hypoglycemia in the sequential add-on therapy group was 1,908 6 50 and 1,953 6 52 mg/day requiring third-party assistance. Down were receiving more than one agent, in combination and insulin therapy titration of medications was allowed and subjects who received initial triple groups, respectively. Similarly, all patients to avoid hypoglycemia. If more than therapy had 0.55% lower HbA1c.From were being treated with a sulfonylurea: three hypoglycemic events per visit on these results we assumed that subjects 58% and 55% of patients were on glicla- two consecutive visits were encoun- with combination therapy (pioglitazone zide and 42% and 45% of patients were on tered, down titration, usually of the sul- plus exenatide) would achieve a lower glimepiride in the combination and insulin fonylurea, was done to lower the risk of HbA1c by $0.55% versus subjects receiv- therapy groups, respectively. During the hypoglycemic events. ing metformin, sulfonylurea, and insulin. study, the sulfonylurea dose could be We computed that 156 subjects per arm adjusted downward if symptomatic/ Data Analysis and Statistical Analysis would provide 90% power to detect a biochemical hypoglycemia was experi- Primary end point was the HbA1c difference 0.55% HbA1c difference between treat- enced. Supplementary Figure 2 dem- between subjects receiving combination ment arms at a , 0.05. onstrates that the reduction in the therapy and those receiving insulin ther- sulfonylurea dose was comparable in apy. Because all patients (N = 231) in the RESULTS both treatment groups. study completed at least 6 months of fol- The study screened 296 patients and ran- The mean follow-up duration was low-up, the full effect of therapy on the domized 251 eligible patients, of whom 12.4 6 0.5 and 12.1 6 0.5 months in HbA1c in each treatment arm was achieved. Therefore, the values of HbA available at 1c — thelastfollow-upvisit(range6–18 months) Table 1 Baseline characteristics of study participants were used for the final analysis. Sec- Combination therapy Insulin therapy n =123 n =108 P value ondary end points included 1) percentage Age (years) 52 6 1526 1NS of subjects achieving HbA1c ,6.5% (48 mmol/mol) and ,7.0% (53 mmol/mol); Male sex (%) 40 37 NS 2) decrease in FPG concentration; 3) BMI (kg/m2)31.16 0.5 30.5 6 0.5 NS change in body weight; and 4)rateof Diabetes duration (years) 10.5 6 0.5 10.9 6 0.5 NS hypoglycemic events. HbA1c (%) 10.0 6 0.6 10.0 6 0.5 NS

The mean value of pre- and postmeal HbA1c (mmol/mol) 86 6 5.2 86 6 4.8 home blood glucose measurements at the FPG (mg/dL) 231 6 82376 7NS last follow-up (range 6–18 months) was Ethnicity (%) used for analysis. Postprandial plasma glu- Qataris 48 39 NS cose excursion was calculated as the Non-Qatari Arabs 26 29 NS incremental area above the FPG con- Asian Indians 17 25 NS Others 9 7 NS centration according to the trapezoid rule. Overall frequency of hypoglycemia was Background therapy Metformin, mg (% of patients) 1,908 (100) 1,953 (100) NS calculated as total number of hypoglyce- Gliclazide, mg (% of patients) 101 (58) 106 (55) NS mic events divided by number of patient- Glimepiride, mg (% of patients) 7.4 (42) 7.2 (45) NS years of follow-up in each arm. The NS, not signiﬁcant. percentage of subjects experiencing 328 Exenatide Plus Pioglitazone vs. Insulin in T2DM Diabetes Care Volume 40, March 2017

the combination therapy and insulin effective in lowering the HbA1c in the therapy groups, respectively. three ethnic groups in the study (Qatari The mean insulin dose in subjects re- nationals, non-Qatari Arabs, and Indians) ceiving insulin therapy is shown in (Supplementary Fig. 4). The efficacy of Supplementary Fig. 3. The mean glargine combination therapy in lowering HbA1c dose was 40 6 1, 47 6 1, and 47 6 2 was independent of age, sex, BMI, or di- units/day at 6, 12, and 18 months, re- abetes duration. spectively, and the mean prandial insulin (aspart) dose was 28 6 1, 40 6 2, and Fasting and Postprandial Glucose 49 6 2 units/day, respectively. Thus, Baseline FPG was similar in the combi- patients in the insulin therapy group re- nation and insulin therapy groups 6 6 ceived an average of ;1.1 units/kg of (224 7 vs. 234 7 mg/dL, respec- fi insulin daily. tively; P = not signi cant) and decreased rapidly after therapy started (Fig. 2A). ThedecreaseinHbA was similar in Primary Outcome 1c The primary outcome was the HbA dif- both groups (Fig. 2A), and at 6 months, 1c 6 6 ference between the two treatment arms the FPG was 107 2vs.105 3mg/dL (P = not significant) in the combination (Fig. 1A). Baseline HbA1c was identical in both groups (10.0% [86 mmol/mol]) and and insulin therapy groups, respectively. progressively decreased in both treat- Figure 2B depicts the postprandial ment groups during the first 6 months. plasma glucose excursion curves in subjects in the combination and insulin However,thedecreaseinHbA1c was greater in subjects receiving combination therapy groups. Subjects in the combi- therapy. At 6 months, there was a statis- nation therapy group experienced a lower postmeal plasma glucose concentra- tically significant difference in the HbA1c (0.7%, P , 0.0001) between the combi- tion. The mean incremental area under the nation therapy (6.7% [50 mmol/mol]) and curve for the plasma glucose concentration insulin therapy (7.4% [57 mmol/mol]) during the day was 28% lower in subjects 6 groups. After 6 months, HbA continued receiving combination therapy (410 1c 6 to decrease in both groups; however, the 17 vs. 536 23mg/dL/hintheinsulin therapy group; P , 0.01). Further, the difference in HbA1c continued to widen with time (0.9 at 12 months and 1% at plasma glucose concentration after break- fi 18 months, both P , 0.0001). fast and dinner was signi cantly lower in More subjects receiving combination subjects receiving combination therapy compared with insulin therapy (Fig. 2B). therapy achieved the ADA treatment Figure 1—Effect of combination therapy and , insulin therapy on glycemic control. A:The goal (HbA1c 7.0% [53 mmol/mol]) ver- Body Weight sus those receiving insulin therapy (83% change over time in the mean HbA1c is The mean body weight increased in both shown in subjects receiving combination vs. 53%, P = 0.003) (Fig. 1B). Similarly, groups. However, subjects in the combi- therapy or insulin therapy. B: The percent- , more patients achieved HbA1c 6.5 nation therapy group experienced half age of subjects is shown in each treatment , (48 mmol/mol) in the combination ther- as much weight gain as subjects in the group achieving HbA1c 7.0% (53 mmol/mol) , and 6.5% (48 mmol/mol). C: The change in apy group (50% vs. 13%, P 0.0001). insulin therapy group (2.1 6 1.1 vs. To examine the efficacy of combination HbA1c over time is shown in subjects in the 4.2 6 1.0 kg, P , 0.0001) (Fig. 3). combination therapy group stratified by base- therapy on glycemic control in relation to line HbA1c into the upper 50% (n = 62) and baseline HbA1c, we divided subjects into Adverse Events lower 50% (n =61).*P , 0.0001; **P = 0.001. two equal groups by their starting HbA1c. In general, both treatments were well tol- Subjects in the upper half (HbA1c .9.5% erated, and a small number (,2%) of pa- [80 mmol/mol], n = 62) had a mean base- tients dropped out of the study because line HbA1c of 11.3 6 0.2% (100 mmol/mol) of adverse events. At least one adverse 6.6 events per patient-year, P , 0.0001). All compared with a mean HbA1c of 8.5 6 event occurred in ;90% of patients in hypoglycemic events were mild, and the 0.1% (69 mmol/mol) in subjects in the lower both groups (Supplementary Table 1); only episode of severe hypoglycemia oc- half (HbA1c 7.5–9.5% [58–80 mmol/mol], most adverse events were mild and un- curredinasubjectintheinsulintherapy n = 61). Figure 2C demonstrates that the related to study treatment. Hypoglycemia group. combination therapy was equally effec- was the most common adverse events re- Two subjects in the combination ther- tive in lowering the HbA1c in both treat- lated to study treatments, being reported apy group experienced a local reaction at ment groups independent of baseline by 91% and 66% of participants receiving the injection site. Subjects receiving com- HbA1c. After 3 months, the difference in insulin therapy and combination therapy, bination therapy experienced more fre- HbA1c in subjects in the upper half was respectively. The overall frequency of hy- quent gastrointestinal adverse effects not statistically different from the mean poglycemic events was approximately and ankle edema. Nausea occurred in HbA1c in subjects in the lower half. Simi- threefold greater in the insulin therapy 31% of combination therapy subjects at larly, combination therapy was equally versus combination therapy group (2.3 vs. the initiation of exenatide therapy; the care.diabetesjournals.org Abdul-Ghani and Associates 329

(CVD) events were in the insulin therapy 15% [80–140 mmol/mol], mean 11.3% group. No CVD events were recorded in [100 mmol/mol]) in the combination ther- the combination therapy group. The apy group decreased at 4 months to number of subjects who withdrew be- the same level as patients with base- cause of adverse events was small: four line HbA1c between7.5and9.5%(58 receiving combination therapy and none and 80 mmol/mol) (mean HbA1c 8.5% receiving insulin therapy. [69 mmol/mol]); there was no difference in the mean HbA1c between groups after CONCLUSIONS 3months(Fig.1C). These results demon- The major new finding of the current strate that the combination of once- study is that combination therapy with weekly exenatide plus pioglitazone, pioglitazone plus exenatide produces a antidiabetic agents that have been robust reduction in the HbA1c in patients shown in previous studies to correct the with poorly controlled (mean HbA1c 10% underlying core defects responsible for the [86 mmol/mol]) long-standing (mean development of hyperglycemia in T2DM duration of 10 years) T2DM treated (i.e., b-cell dysfunction and insulin resis- with maximal/nearly maximal doses of tance) (12–20), reduces the mean plasma metformin plus a sulfonylurea. glucose concentration close to the normal It should be noted that subjects in both range, regardless of the starting HbA1c, treatment arms, despite very poor glyce- and with a low risk of hypoglycemia. mic control at baseline (HbA1c 10% Subjects receiving combination ther- [86 mmol/mol]), generally experi- apy experienced half of the weight gain Figure 2—Effect of combination therapy and enced very good glycemic control. As an- compared with subjects receiving insulin insulin therapy on FPG concentration (Conc.) ticipated, insulin therapy, which is very therapy (2.1 vs. 4.2 kg) (Fig. 3). Although (A) and postprandial plasma glucose concen- powerful in lowering the plasma glucose pioglitazone therapy resulted in weight tration (B) in The Qatar Study. AB, after breakfast; AL, after lunch; AS, after supper; BL, concentration, produced a robust reduc- gain, 32% of patients receiving combina- before lunch; BS, before supper; BT, bedtime. tion in mean HbA1c (Δ = 2.9%), and half of tion therapy experienced weight loss the subjects receiving insulin therapy compared with only 13% (P , 0.01) in achieved the ADA treatment goal of the insulin treatment group. This most HbA1c ,7.0% (53 mmol/mol) without sig- likely is explained by the concomitant nausea was mild and subsided after nificant hypoglycemia. The mean FPG use of exenatide, which is well docu- – 2 3 months. Only two patients discontin- concentration in insulin-treated sub- mented to suppress appetite and cause ued treatment as a result of nausea. jects reached the target (110 mg/dL) at weight loss (20,21). It also should be The incidence of peripheral edema was ;2 months, confirming rigorous insulin in- pointed out that with pioglitazone, the low, at 3.4% vs. 9.3% in insulin and com- tensification in the insulin-treated group. greater the weight gain, the greater are bination therapy, respectively. Peripheral However, the glycemic goal (HbA1c,7.0% the improvements in HbA1c, insulin sensi- edema was mild and easily controlled [53 mmol/mol]) was achieved in only half tivity, and b-cell function (17). with the addition of a distally acting di- of the patients because hypoglycemia pre- Both treatments were well tolerated, uretic. Two patients discontinued combina- cluded further escalation of the insulin and few subjects (,3% in combination tion therapy because of peripheral edema. dose. Subjects receiving combination ther- therapy; 0% in insulin therapy) withdrew There were no cases of congestive heart apy experienced a more robust reduction because of adverse events. Most of the failure. Five cardiovascular events occurred in mean HbA1c (Δ = 3.9%). Because neither adverse events were mild and unrelated (4 myocardial infarctions, 1 of which was pioglitazone nor exenatide are associated to study treatment. Hypoglycemia was fatal; and 1 cerebrovascular accident). The with an increased risk of hypoglycemia, the most common adverse event; the ab- fi ve subjects with cardiovascular disease subjects receiving combination therapy solute hypoglycemia rate was relatively experienced a threefold lower risk of hy- low (Supplementary Table 1) compared poglycemia despite an HbA1c that was 1% with other studies that have used inten- lower compared with the insulin therapy sive insulin therapy in T2DM (23–25). The group. Of note, most of the hypoglycemic relatively low rate of peripheral edema events in subjects receiving combination (Supplementary Table 1) with combina- therapy were related to the use of sulfonyl- tion therapy most likely is explained by ureas, because down titration or stopping the lower dose of pioglitazone (30 mg/day) the sulfonylurea in these patients elimi- used in the current study. The natri- nated most of the hypoglycemic events. uretic effect of exenatide also could have The combination of pioglitazone plus reduced the incidence of peripheral exenatide was very effective in lower- edema. Most of edema that occurred ing the HbA independent of ethnicity, was mild and easily controlled with the Figure 3—Effect of combination therapy and 1c insulin therapy on body weight. BMI, sex, or baseline HbA1c. Remarkably, addition of a distally acting diuretic to the HbA1c in subjects in the upper half the treatment regimen. Only two sub- of the HbA1c range (baseline HbA1c 9.5– jects discontinued combination therapy 330 Exenatide Plus Pioglitazone vs. Insulin in T2DM Diabetes Care Volume 40, March 2017

because of ankle edema. No congestive patients with poorly controlled, long- a patient-centered approach: position statement heart failure or fractures occurred in ei- standing T2DM on metformin plus a sul- of the American Diabetes Association (ADA) and ther group. fonylurea. Continued follow-up will be the European Association for the Study of Diabe- tes (EASD). Diabetes Care 2012;35:1364–1379 Although the study was not powered required to ascertain how long the ben- 7. DeFronzo RA, Goodman AM; The Multicenter to detect the effect of therapy on car- eficial effects of combination therapy Metformin Study Group. Efficacy of metformin in diovascular events, five cardiovascular are maintained. patients with non-insulin-dependent diabetes events, one fatal, occurred in the insulin mellitus. N Engl J Med 1995;333:541–549 therapy group. Because of the small sam- 8. Rossetti L, DeFronzo RA, Gherzi R, et al. Ef- fect of metformin treatment on insulin action in ple size in the current study and short Acknowledgments. The authors thank Evette diabetic rats: in vivo and in vitro correlations. follow-up duration, it is impossible to de- Ibrahim, RN, Sanaa Mansy, RN, and Huda Mejri, Metabolism 1990;39:425–435 termine whether this result was due to RN, (Hamad General Hospital) for their excellent 9. Kahn SE, Haffner SM, Heise MA, et al.; ADOPT care of patients and Huda Esam, Mariam Study Group. Glycemic durability of rosiglita- chance or to a potential protective cardio- Al-Malaheem, and Kirollos Magdi (Hamad Gen- vascular effect of pioglitazone (26) or zone, metformin, or glyburide monotherapy. eral Hospital) for technical and logistic support N Engl J Med 2006;355:2427–2443 GLP-1 RA (27), both of which have been throughout the study. The authors also thank 10. United Kingdom Prospective Diabetes shown to reduce CVD events, but longer the pharmacy team led by Dr. Enas Abdoun Study (UKPDS). United Kingdom Prospective Di- monitoring will be required to see the and the team of diabetes educators at Hamad abetes Study (UKPDS). 13: relative efficacy of General Hospital for dispensing the study med- CVD benefit. A much larger and longer randomly allocated diet, sulphonylurea, insulin, ications and for training patients on the use of or metformin in patients with newly diagnosed cardiovascular outcome study will be re- insulin and BYDUREON pens, respectively. non-insulin dependent diabetes followed for quired to answer this question. Funding. This study was supported by a Qatar three years. BMJ 1995;310:83–88 Foundation grant NPRP 5-273-3-079. R.A.D.’s The current study has several limi- 11. Seaquist ER, Anderson J, Childs B, et al. Hy- salary is paid in part by the South Texas Vet- poglycemia and diabetes: a report of a work- tations. The therapies used (weekly erans Health Care System. group of the American Diabetes Association exenatide and multiple daily insulin in- R.A.D. is on the advisory Duality of Interest. and the Endocrine Society. Diabetes Care boards of AstraZeneca, Novo Nordisk, Janssen, jections) prohibited the study from 2013;36:1384–1395 Intarcia, and Boehringer-Ingelheim; receives re- being blinded. Because study was per- 12. Bays H, Mandarino L, DeFronzo RA. Role of search support from Bristol-Myers Squibb, formed in a single center, it included a the adipocyte, free fatty acids, and ectopic fat in Boehringer-Ingelheim, Takeda, and AstraZeneca; pathogenesis of type 2 diabetes mellitus: perox- relatively small number of participants, and is on the speaker’s bureaus of Novo Nordisk isomal proliferator-activated receptor agonists primarily of Arab origin (;70% of partic- and AstraZeneca. AstraZeneca provided the ex- provide a rational therapeutic approach. J Clin enatide. No other potential conflicts of interest ipants). Thus, a larger multiethnic study Endocrinol Metab 2004;89:463–478 relevant to this article were reported. is warranted to examine the generaliz- 13. Rasouli N, Kern PA, Reece EA, Elbein SC. Author Contributions. M.A.-G. designed the ability of this novel treatment approach Effects of pioglitazone and metformin on beta- study, wrote the protocol, contributed to data cell function in nondiabetic subjects at high risk in individuals with poorly controlled, generation and data analysis, and wrote the man- for type 2 diabetes. Am J Physiol Endocrinol long-standing T2DM. Further, longer uscript.O.M.,A.M.,J.A.,andC.T.generatedthe Metab 2007;292:E359–E365 data. R.A.D., M.Z., and A.J. reviewed and revised follow-up is required to determine the 14. Xiang AH, Peters RK, Kjos SL, et al. Effect of the manuscript. M.A.-G. is the guarantor of this durability of glycemic control achieved pioglitazone on pancreatic beta-cell function and work and, as such, had full access to all the data in in the combination therapy group. diabetes risk in Hispanic women with prior gesta- the study and takes responsibility for the integrity tional diabetes. Diabetes 2006;55:517–522 Whether the HbA1c difference (6.1% of the data and the accuracy of the data analysis. [43 mmol/mol] in combination therapy 15. Jin J, Yu Y, Yu H, Wang C, Zhang X. Effects of pioglitazone on beta-cell function in metabolic vs. 7.1% [54 mmol/mol] in insulin ther- References syndrome patients with impaired glucose toler- apy) in this range of HbA1c will be trans- 1. DeFronzo RA. Banting Lecture. From the tri- ance. Diabetes Res Clin Pract 2006;74:233–241 lated into greater prevention of diabetic umvirate to the ominous octet: a new paradigm 16. DeFronzo RA, Tripathy D, Schwenke DC, microvascular complications also re- for the treatment of type 2 diabetes mellitus. et al.; ACT NOW Study. Pioglitazone for diabetes mains to be determined. Diabetes 2009;58:773–795 prevention in impaired glucose tolerance. N Engl J Med 2011;364:1104–1115 Nonetheless, the present results are 2. The Diabetes Control and Complications Trial Research Group. The effect of intensive treatment 17. Gastaldelli A, Ferrannini E, Miyazaki Y, impressive and, contrary to standard of diabetes on the development and progression of Matsuda M, Mari A, DeFronzo RA. Thiazolidine- dogma, demonstrate that even in indi- long-term complications in insulin-dependent dia- diones improve beta-cell function in type 2 di- viduals with very poorly controlled, betes mellitus. N Engl J Med 1993;329:977–986 abetic patients. Am J Physiol Endocrinol Metab – long-standing T2DM, combination ther- 3. UK Prospective Diabetes Study (UKPDS) 2007;292:E871 E883 18. Mazzone T, Meyer PM, Feinstein SB, et al. Ef- apy with a GLP-1 RA plus pioglitazone Group. Intensive blood-glucose control with sul- phonylureas or insulin compared with conven- fect of pioglitazone compared with glimepiride on can achieve nearly normal/normal tional treatment and risk of complications in carotid intima-media thickness in type 2 diabetes: a HbA1c levels. We believe that the robust patients with type 2 diabetes (UKPDS 33). Lan- randomized trial. JAMA 2006;296:2572–2581 resultsobservedinthecurrentstudy cet 1998;352:837–853 19. Hanefeld M, Pfutzner¨ A, Forst T, Lubben¨ G. should foster a larger multiethnic study 4. Nakagami T, Kawahara R, Hori S, Omori Y. Glycemic control and treatment failure with pioglitazone versus glibenclamide in type 2 di- with longer duration to validate the Glycemic control and prevention of retinopathy in Japanese NIDDM patients. A 10-year fol- abetes mellitus: a 42-month, open-label, obser- findings in the current study and exam- low-up study. Diabetes Care 1997;20:621–622 vational, primary care study. Curr Med Res Opin ine their benefit on diabetic microvascu- 5. Klein R, Klein BE, Moss SE. Relation of glycemic 2006;22:1211–1215 lar and macrovascular complications. control to diabetic microvascular complications in 20. Drucker DJ. The biology of incretin hor- In summary, the present results dem- diabetes mellitus. Ann Intern Med 1996;124:90–96 mones. Cell Metab 2006;3:153–165 6. Inzucchi SE, Bergenstal RM, Buse JB, et al.; 21. Bunck MC, Corner´ A, Eliasson B, et al. Effects onstrate that combination therapy with American Diabetes Association (ADA); European of exenatide on measures of b-cell function after pioglitazone plus exenatide is a very ef- Association for the Study of Diabetes (EASD). 3 years in metformin-treated patients with type 2 fective and safe therapeutic option in Management of hyperglycemia in type 2 diabetes: diabetes. Diabetes Care 2011;34:2041–2047 care.diabetesjournals.org Abdul-Ghani and Associates 331

22. Abdul-Ghani MA, Puckett C, Triplitt C, et al. 24. Fritsche A, Larbig M, Owens D, Haring¨ HU; 26. Dormandy JA, Charbonnel B, Eckland DJ, Initial combination therapy with metformin, pio- GINGER study group. Comparison between a et al.; PROactive Investigators. Secondary pre- glitazone and exenatide is more effective than basal-bolus and a premixed insulin regimen in indi- vention of macrovascular events in patients sequential add-on therapy in subjects with new- viduals with type 2 diabetes-results of the GINGER with type 2 diabetes in the PROactive Study onset diabetes. Results from the Efﬁcacy and Du- study. Diabetes Obes Metab 2010;12:115–123 (PROspective pioglitAzone Clinical Trial In mac- rability of Initial Combination Therapy for Type 2 25. Bowering K, Reed VA, Felicio JS, Landry J, Ji roVascular Events): a randomised controlled tri- Diabetes (EDICT): a randomized trial. Diabetes L, Oliveira J. A study comparing insulin lispro mix al. Lancet 2005;366:1279–1289 Obes Metab 2015;17:268–275 25 with glargine plus lispro therapy in patients 27. Marso SP, Daniels GH, Brown-Frandsen K, 23. Lingvay I, Legendre JL, Kaloyanova PF, Zhang S, with type 2 diabetes who have inadequate gly- et al.; LEADER Steering Committee; LEADER Trial Adams-Huet B, Raskin P. Insulin-based versus triple caemic control on oral anti-hyperglycaemic Investigators. Liraglutide and cardiovascular oral therapy for newly diagnosed type 2 diabetes: medication: results of the PARADIGM study. Di- outcomes in type 2 diabetes. N Engl J Med which is better? Diabetes Care 2009;32:1789–1795 abet Med 2012;29:e263–e272 2016;375:311–322