MRIGTEAIS RG N REGIMENS AND DRUGS THERAPIES: EMERGING

Diabetes Care 1

Christopher K. Rayner,1,2 Linda E. Watson,1 Effects of Sustained Treatment Liza K. Phillips,1,3 Kylie Lange,1 Michelle J. Bound,1 Jacqueline Grivell,1 With Lixisenatide on Gastric Tongzhi Wu,1,3 Karen L. Jones,1,3 Michael Horowitz,1,3 Ele Ferrannini,4 Emptying and Postprandial Domenico Trico,` 5 Silvia Frascerra,6 Glucose Metabolism in Type 2 Andrea Mari,7 and Andrea Natali6 Diabetes: A Randomized Controlled Trial https://doi.org/10.2337/dc20-0190

OBJECTIVE Tachyphylaxis for slowing of gastric emptying is seen with continuous exposure to glucagon-like peptide 1 (GLP-1). We therefore aimed to establish whether prolonged use of a “short-acting” GLP-1 receptor agonist, lixisenatide, achieves sustained slowing of gastric emptying and reduction in postprandial glycemia.

RESEARCH DESIGN AND METHODS 1Centre of Research Excellence for Translating A total of 30 patients with metformin-treated type 2 diabetes underwent assessment Nutritional Science to Good Health, University of Adelaide, Adelaide, Australia of gastric emptying (scintigraphy) and glucose metabolism (dual tracer technique) 2Department of Gastroenterology and Hepatol- aftera75-gglucosedrink,beforeandafter8weeks’treatmentwithlixisenatide(20mg ogy, Royal Adelaide Hospital, Adelaide, Australia subcutaneously daily) or placebo, in a double-blind randomized parallel design. 3Endocrine and Metabolic Unit, Royal Adelaide Hospital, Adelaide, Australia RESULTS 4CNR Institute of Clinical Physiology, Pisa, Italy 5Department of Surgical, Medical and Molecular Gastric retention of the glucose drink was markedly increased after lixisenatide Pathology and Critical Care Medicine, University versus placebo (ratio of adjusted geometric means for area under curve [AUC] over of Pisa, Pisa, Italy 240min of 2.19[95%CI 1.82, 2.64;P< 0.001]), associatedwith substantial reductions 6Department of Clinical and Experimental Med- in the rate of systemic appearance of oral glucose (P < 0.001) and incremental AUC icine, University of Pisa, Pisa, Italy 7CNR Institute of Neuroscience, Padua, Italy for blood glucose (P < 0.001). Lixisenatide suppressed both glucagon (P 5 0.003) and P 5 Corresponding author: Christopher K. Rayner, insulin ( 0.032), but not endogenous glucose production, over 120 min after oral [email protected] glucose intake. Postprandial glucose lowering over 240 min was strongly related to Received 27 January 2020 and accepted 28 April the magnitude of slowing of gastric emptying by lixisenatide (r 520.74, P 5 0.002) 2020 and to the baseline rate of emptying (r 5 0.52, P 5 0.048) but unrelated to b-cell Clinical trial reg. no. ACTRN12616001059459, function (assessed by b-cell glucose sensitivity). www.anzctr.org.au This article contains supplementary material online CONCLUSIONS at https://doi.org/10.2337/figshare.12245516. Eight weeks’ treatment with lixisenatide is associated with sustained slowing of © 2020 by the American Diabetes Association. gastric emptying and marked reductions in postprandial glycemia and appearance Readers may use this article as long as the work is properly cited, the use is educational and not for of ingested glucose. Short-acting GLP-1 receptor agonists therefore potentially fi fi pro t, and the work is not altered. More infor- represent an effective long-term therapy for speci cally targeting postprandial mation is available at https://www.diabetesjournals glucose excursions. .org/content/license. Diabetes Care Publish Ahead of Print, published online May 29, 2020 2 Gastric Emptying With Sustained Lixisenatide Diabetes Care

Early intervention to achieve good gly- The evolution of type 2 diabetes is Study Design cemic control in type 2 diabetes can characterized by progressive b-cell After enrolment, participants were ran- reduce the risk of micro- and potentially failure, which entails a loss of insulin domized to treatment with lixisenatide macrovascular complications but entails secretory capacity and, eventually, a re- or placebo in a double-blind parallel risks, particularly for insulin therapy, of duction in b-cell mass (16). A therapy that design. Patients attended our Clinical weight gain and hypoglycemia (1). Re- targets postprandial glycemia, without Research Facility for a baseline study cent guidelines recommend an individ- the requirement to stimulate insulin se- where a glucose drink was administered ualizedapproachtotheselectionof cretion, is an attractive therapeutic op- (day 0) before commencing treatment second-line pharmacological treatment tion, and there is recent evidence that with daily subcutaneous injections from after metformin (2). In the majority of lixisenatide maintains its efficacy for day 1 to day 56. They attended the type 2 patients in the community, who lowering HbA1c across the spectrum of laboratory for a second study on the have relatively modest elevation of gly- b-cell function in patients with type 2 last day of treatment (day 56), when fi cated hemoglobin (HbA1c) (3), postpran- diabetes (17). they received their nal dose of lixise- dial rather than fasting blood glucose We aimed to evaluate the effects of natide or placebo 30 min before the plays an important role in overall glyce- sustained use of lixisenatide on gastric glucose test drink. mic control (4), which is not surprising emptying, using the most accurate tech- On each of the two study days (days given that humans are predominantly in nique of scintigraphy, and on postpran- 0 and 56), patients attended our facility the postprandial state. It is now apparent dial glucose metabolism, using a dual in the morning (0800 h) after an over- that postprandial glycemia must be spe- glucose tracer technique (18), in patients night fast, having consumed a standard- cifically targeted in patients with diabe- with relatively well-controlled type 2 di- ized beef lasagne meal the previous tes, particularly in those who have only abetes treated with metformin alone. We evening. An intravenous (IV) cannula moderately elevated HbA1c with minimal hypothesized that, as a short-acting GLP-1 wasinsertedineachforearm:onefor fasting hyperglycemia (5). receptor agonist, lixisenatide would have blood sampling and the other for IV The gut-derived incretin hormone, glu- persistingeffectstoslowgastricemptying infusion of glucose tracer (bolus of 21 2 cagon-like peptide 1 (GLP-1), has several with sustained use and that this would 28 mmol z kg 6,6-[ H2]glucose, fol- actions that lower glycemia, including be a dominant mechanism by which it lowed by continuous infusion at a 2 2 slowing of gastric emptying, glucose- achieves postprandial blood glucose rate of 0.28 mmol z min 1 z kg 1 dependent stimulation of insulin, and sup- control. from t 52180 until t 5 240 min). At pression of glucagon (6–8), all of which t 525 min, a 300-mL drink was given potentially impact on the systemic ap- RESEARCH DESIGN AND METHODS while the subject sat against a g camera, pearance of orally ingested carbohydrate Subjects consisting of 75-g glucose, labeled with and the rate of endogenous glucose We recruited, by advertisement, patients 20 MBq 99mTc-calcium phytate, and also production. The rate at which nutrients with a history of type 2 diabetes of $2 containing 1.5 g [U-13C]glucose. The drink empty from the stomach is a particularly years’ duration, treated with metformin was consumed within 5 min, and scinti- important determinant of postprandial for $3 months. Responders who met graphic imaging continued for 240 min, glycemia in health and type 2 diabetes (9), eligibility criteria attended our facility with correction of data for subject move- and pharmacological agents that slow gas- for a screening visit, when fasting blood ment, radionuclide decay, and g-ray tric emptying result in a lowering of the samples were collected for blood picture, attenuation (21). Gastric retention (ex- postprandial blood glucose excursion (10). iron studies, biochemistry, and HbA1c; pressed as % of the maximum content A number of GLP-1 receptor agonists autonomic function was evaluated by of the total stomach) was calculated at have been developed for the treatment standardized cardiovascular tests (19); 15-min intervals for the first 120 min of type 2 diabetes; these differ in being and gastrointestinal symptoms were andthenhourlyuntilt 5 240 min. “short acting” (e.g., lixisenatide or ex- evaluated by questionnaire (20). Those Venous blood was sampled at frequent 21 enatide BD) or “long acting” (e.g., liraglu- with HbA1c ,6.5% (48 mmol z mol ) intervals for measurement of blood tide or exenatide QW). There is recent or .9.0% (75 mmol z mol21), estimated glucose and plasma concentrations of evidence that continuous exposure to glomerular filtration rate ,30 mL z min21, glucosetracers,insulin,C-peptide,and GLP-1is associated withtachyphylaxis for definite autonomic neuropathy, elevated glucagon. A fasting sample was also its effects on gastric emptying (11,12). liver enzymes, hospital admission for car- collected on day 56 to measure HbA1c. Consistent with this concept, postpran- diovascular disease within the previous dial glycemia appears to be better con- 6 months, previous exposure to GLP-1 Intervention trolled after several weeks’ treatment receptor agonists, or prominent gastro- On days 1–56 inclusive, each subject self- with short-acting versus long-acting GLP- intestinal symptoms or using medications administered lixisenatideor matchingpla- 1 receptor agonists, although typically in known to affect gastrointestinal motility cebo (saline) subcutaneously once daily, such studies, either gastric emptying has were excluded. All participants gave writ- 30 min before breakfast. The dose of not been measured (13) or a suboptimal ten informed consent, and the protocol lixisenatide was stepped up from 5 mg technique has been useddsuch as para- was approved by the Human Research on days 1–7to10mgondays8–14 and cetamol absorption (14) or a stable Ethics Committee of the Royal Adelaide 20 mgondays15–56. Blinding and ran- isotope breath test involving a meal Hospitalandprospectivelyregisteredwith domization were coordinated by the different from that used to assess the the Australian New Zealand Clinical Trials hospital pharmacy, and subjects were glycemic response (15). Registry (ACTRN12616001059459). educated in injection technique at the care.diabetesjournals.org Rayner and Associates 3

day0visit.Inadditiontothestudyvisitson endogenous glucose production, as well coefficient (b),and the partial correlation day 0 and day 56, subjects attended our as changes in fasting blood glucose (eval- (rpartial)arepresented;thelastrepresents facility onday 7 tocollect their medication uated from the sample immediately prior the correlation between the indepen- supply for the following week and on day to the glucose drink) and HbA1c.Param- dent variable and outcome after removal 14 toreceive suppliesforthe remainderof eters of b-cell function (fasting and total of the effect of the other independent theintervention.Ateachvisit,participants insulin secretion, glucose sensitivity, and variable. were weighed and completed a gastro- rate sensitivity and potentiation) were Data are reported as means and SDs or intestinal symptom questionnaire (20). resolved from mathematical modeling 95% CIs or as median and interquartile They also completed questionnaires at of oral glucose tolerance data as pre- ranges (IQRs) if not normally distributed. days 28, 42, and 56. Compliance was viously described (24). Additional ex- Statistical significance was set as P , 0.05. monitored at each of these visits by ploratory end points were changes in recording used and unused medication, fasting blood glucose and plasma in- RESULTS and subjects also recorded each dose of sulin, C-peptide, and glucagon concen- Fifty-three patients with type 2 diabetes study medication in a daily diary. trations and body weight. attended for a screening visit, of whom Analysis was undertaken on a per- 33 were randomized (Supplementary Fig. Measurements protocol basis. The effect of lixisenatide 1). Of the latter, 30 patients commenced Blood glucose concentrations were de- was analyzed using ANCOVA, with fixed treatment, all of whom completed the terminedbyportableglucometer(Optium effects for treatment group (placebo or study and were included in the analysis Xceed; Abbott Laboratories, Chicago, IL) lixisenatide) and the baseline value of (9 female and 21 male, mean (SD) age (mean of two measurements at each time the outcome parameter. The baseline- 67.1 (6.0) years and BMI 32.1 (5.1) kg/m2, point). The remainder of each blood sample adjusted treatment difference at day 56, median duration of known diabetes 5.5 was placed into ice-chilled EDTA-treated 95% CI, and P value for the treatment years (IQR 3.0–10.0), and mean HbA1c tubes, and plasma was separated by effect are reported. Residuals for each 7.1% (0.6%) (53 [6.6] mmol/mol)). One centrifugation and stored at 280°C until model were examined for normality, had evidence of peripheral neuropathy, assayed. Insulin was measured by ELISA linearity, and constant variance. Gastric and two had a history of ischemic heart (10-1113; Mercodia, Uppsala, Sweden), emptyingAUCwastransformedusing the disease, but no others had evidence of with sensitivity 1.0 mU z L21 and intra- and natural log due to nonconstant variance, micro- or macrovascular complications of interassay coefficients of variation (CVs) and the treatment effect is reported as diabetes. The characteristics of patients 2.9% and 6.7%, respectively. C-peptide the ratio of the geometric means. A ratio in each of the two groups (placebo n 5 was measured by ELISA (10-1136-01; of 1 indicates equal means in the two 15, lixisenatide n 5 15) are presented Mercodia), with sensitivity 15 pmol z groups, while values .1 indicate a higher (Supplementary Table 1). The oral glu- L21 and intra- and interassay CVs 7.7% mean for lixisenatide than for placebo cose tracer was inadvertently omitted and 3.7%. Glucagon was measured by and values ,1 indicate a lower mean for from the glucose drink in one patient in radioimmunoassay (GL-32 K; Millipore, lixisenatide than placebo. Change over the placebo group, so glucose tracer data Billerica, MA), with sensitivity 20 pg z time was calculated as day 56 2 day 0, so were only available for 14 placebo-treated mL21 and intra- and interassay CVs 6.4% negative values indicate a decrease and patients. and3.2%.Glucosetracersweremeasured positive values an increase over time. The intervention was well tolerated, using gas chromatography–mass spec- Differences between groups were calcu- andcompliancewasexcellent;patients in trometry and glucose fluxes calculated lated as lixisenatide 2 placebo, so neg- both the placebo and lixisenatide groups from the time course of the plasma tracer- ative values indicate that lixisenatide took a median 98.2% (IQR 98.2–100) of 2 to-tracee ratio of 6,6-[ H2]glucose and had a lower mean than placebo had and scheduled doses. Seven patients treated [U-13C]glucose (22). positive values that lixisenatide had a with placebo reported adverse effects higher mean. (including abdominal pain in three, con- Statistical Analysis Within the lixisenatide group, relation- stipation in one, diarrhea in one, and The prespecified primary outcome mea- ships between changes in postprandial vomiting in one), while eight patients sure was change in gastric half emptying glucose, gastric retention, and average treated with lixisenatide reported ad- time from baseline to day 56 for lixisena- oral glucose Ra are presented as re- verse effects (including abdominal pain tide versus placebo. A sample size of stricted cubic splines (with three knots in one, constipation in two, diarrhea in 30 patients (15 in each group) was at 0.1, 0.5, and 0.9). When linearity five, vomiting in one, nausea in four, calculated to provide .90% power to appeared reasonable, the Pearson cor- bloating in two, and acid reflux in two). detect a 50% increase in gastric half relation was calculated. Multiple linear All adverse effects were transient and emptying time in the lixisenatide group regression was used to determine the resolved spontaneously. Gastrointesti- (23).Gastricemptyingwasalsoevaluated independent associations of change in nal symptom questionnaires adminis- by area under the curve (AUC) for re- postprandial glycemia with change in teredatscreeningandondays7,14, tention over 0–240 and 0–120 min, gastric retention and change in b-cell 28, 42, and 56 revealed low scores in calculated using the trapezoidal rule. glucose sensitivity. The regression model both groups for poor appetite, nausea, Secondary end points included incre- was assessed for normality, linearity, fullness, abdominal discomfort, vomiting, mental AUC (iAUC) for blood glucose, constant variance, and multicollinearity. abdominal pain, dysphagia, heartburn, and plasma insulin, C-peptide, gluca- The unstandardized regression coeffi- and acid regurgitation (data not shown). gon, Ra of orally ingested glucose, and cient (b), the standardized regression Mean body weight declined slightly in 4 Gastric Emptying With Sustained Lixisenatide Diabetes Care

both groups, but there was no significant curve in the lixisenatide group on day closely related to the slowing of gastric adjusted group difference at day 56 (P 5 56, such that the iAUC (0–120 min) emptying with lixisenatide. 0.714). showed a significant adjusted reduction The change in postprandial glucose (P 5 0.032), although the difference was iAUC after lixisenatide was also related Gastric Emptying not significant for iAUC (0–240 min) (P 5 togastric retentionAUC at baseline for0– fi The prespeci ed primary outcome mea- 0.209) (Table 1 and Fig. 2D). 240 min (Pearson r 5 0.52, P 5 0.048) sure of gastric half emptying time could and nonsignificantly for 0–120 min (Pear- not be used, as 4 of 15 patients treated Plasma C-Peptide Concentrations son r 5 0.45, P 5 0.093), indicating that with lixisenatide (and none treated with Similar to insulin, fasting C-peptide was patients with more rapid gastric empty- . fl placebo) had 50% retention of the not altered but there was attening of ing at baseline had greater reductions in glucose drink in the stomach at 240 min the postprandial C-peptide curve in the postprandial blood glucose with lixisena- on day 56. With evaluation instead of AUC lixisenatide group on day 56, although tide treatment. – for gastric retention over 0 240 min, the adjusted group differences did not reach The relationship between change in fi – ratio of adjusted geometric means for statistical signi cance for either iAUC (0 postprandial glucose iAUC after lixisena- 5 – lixisenatide and placebo at day 56 was 120 min) (P 0.090) or iAUC (0 240 min) tide and change in b-cell glucose sensi- , 5 2.19 (95% CI 1.82, 2.64; P 0.001), (P 0.519) (Table 1 and Fig. 2E). tivity did not reach statistical significance indicating marked slowing of gastric for 0–240 min (Pearson r 520.40, P 5 emptying following 8 weeks’ treatment Plasma Glucagon Concentrations – 52 Fasting glucagon concentrations were not 0.138) or 0 120 min (Pearson r 0.43, with lixisenatide. Similarly, with evalu- 5 altered with lixisenatide (P 5 0.165 for P 0.109). In the multiple regression ation of AUC for gastric retention over model, the change in gastric retention – adjusted group difference at day 56), but 0 120 min, the ratio of adjusted geo- remainedstronglyassociatedwithchange metric means for lixisenatide and pla- there was suppression of the postprandial in postprandial glucose iAUC (rpartial 5 cebo at day 56 was 1.55 (95% CI 1.37, glucagon curve in the lixisenatide group fi 20.87, b 520.85, P , 0.001 for 0– 1.74; P , 0.001) (Table 1 and Fig. 1). on day 56, with a statistically signi cant adjusted group difference for iAUC (0– 120 min) after adjustment for the change in b-cell glucose sensitivity, whereas the Blood Glucose Concentrations 120 min) (P 5 0.003) but not iAUC (0– Fasting blood glucose declined slightly in 240 min) (P 5 0.151) (Table 1 and Fig. 2F). association with the change in b-cell the lixisenatide group on day 56 (mean glucose sensitivity was not independently 21 b significant (r 520.24, b 520.12, [SD] 20.9 [0.8] mmol z L , ratio of geo- HbA1c and -Cell Function partial P 5 0.40). Finally, the change in post- metric means for differences after treat- HbA1c was reduced after lixisenatide ment 0.89 [95% CI 0.82, 0.97]. There was versus placebo (adjusted group differ- prandial glucose iAUC after lixisenatide 2 2 marked flattening of the postprandial ence at day 56 of 0.48% (CI 0.70, was not related to b-cell glucose sensi- 2 2 z 21 2 5 5 blood glucose curve, such that both the 0.26) ( 5.2 mmol mol [CI 7.7, tivity at baseline (Pearson r 0.25, P 2 , iAUC (0–240 min) and iAUC (0–120 min) 2.8], P 0.001). At day 56, fasting 0.379). insulin secretion rate was significantly showed substantial adjusted group dif- CONCLUSIONS ferences at day 56 (P , 0.001 for each) reduced after lixisenatide versus pla- We have shown that the short-acting (Table 1 and Fig. 2A). cebo, while total insulin secretion over 240 min was not different between the GLP-1 receptor agonist lixisenatide indu- R of Oral Glucose twogroups.b-Cellglucosesensitivity was ces substantial slowing of gastric emp- a tying that persists over several weeks, The Ra of oral glucose was greatly reduced almost doubled following lixisenatide in the lixisenatide group on day 56, such treatment (Supplementary Fig. 2), while associated with marked reductions in that there were large adjusted group rate sensitivity and potentiation were postprandial glycemia and systemic ap- fi pearance of ingested glucose in metformin- differences for average Ra over both 0– not affected signi cantly (Table 1). 240 min and 0–120 min (P , 0.001 for treated patients with type 2 diabetes. A each) (Table 1 and Fig. 2B). Relationships Between Variables short-acting GLP-1 receptor agonist, such The change in postprandial glucose iAUC as lixisenatide, therefore potentially rep- Rate of Endogenous Glucose after 8 weeks’ treatment with lixisena- resents an effective long-term therapy for fi Production tide was strongly related to the change in speci cally targeting postprandial blood Endogenous glucose production was not gastric retention AUC for both 0–240 min glucose excursions. significantly altered after lixisenatide (Pearson r 520.74, P 5 0.002) and 0– We observed that the more lixisena- treatment,suchthattherewerenogroup 120 min (Pearson r 520.89, P , 0.001) tide slowed gastric emptying, the greater differences in the average rate of endog- (Fig. 3), indicating that patients who the lowering of postprandial blood glu- enous glucose production at day 56 for achieved greater slowing of gastric emp- cose, and also that when gastric empty- either 0–240 min (P 5 0.452) or 0– tying with lixisenatide treatment had a ing was more rapid at baseline, the 120 min (P 5 0.152) (Table 1 and Fig. 2C). more substantial reduction in postpran- benefit of lixisenatide in terms of glucose dial glycemia. The change in average Ra lowering was greater. We recently re- Plasma Insulin Concentrations of oral glucose was also related to the ported that patients with relatively un- There was no significant change in fasting change in gastric retention AUC over complicated type 2 diabetes, as a group, insulin (adjusted group difference at day 0–240 min (Pearson r 520.66, P 5 have faster gastric emptying than age- 56 P 5 0.137). However, there was 0.007), indicating that the reduction in matched control subjects (25), which flattening of the postprandial insulin systemic appearance of oral glucose was would make such patients particularly Table 1—Outcomes Rayner and Associates 5 Adjusted group difference Placebo (N 5 15) Lixisenatide (N 5 15) (ratio of geometric means) Day 0 Day 56 Change Day 0 Day 56 Change Mean (95% CI) P Gastric retention AUC 0–240 min (% z min) 8,544 (1,255) 8,394 (1,698) 2150 (1,171) 8,339 (1867) 18,370 (4,858) 10,031 (4,926) 2.19 (1.82, 2.64)^ ,0.001 Gastric retention AUC 0–120 min (% z min) 7,454 (944) 7,120 (1,189) 2334 (891) 7,165 (1,092) 10,770 (1,464) 3,605 (1,804) 1.55 (1.37, 1.74)^ ,0.001 Fasting blood glucose (mmol z L21) 8.3 (1.2) 8.4 (2.0) 0.1 (1.2) 7.6 (1.7) 6.7 (1.1) 20.9 (0.8) 0.89 (0.82, 0.97)^ 0.013 2 Blood glucose iAUC 0–240 min (mmol z L 1 z min) 1,388 (338) 1,375 (293) 214 (288) 1,407 (491) 461 (455) 2947 (414) 2925 (21,156, 2693) ,0.001 2 Blood glucose iAUC 0–120 min (mmol z L 1 z min) 777 (162) 802 (154) 26 (212) 842 (299) 239 (233) 2602 (328) 2575 (2723, 2426) ,0.001 21 21 Average RaO0–240 min (mmol z min z kg ) 10.5 (2.7) 10.6 (1.9) 0.2 (1.9) 10.9 (2.6) 5.2 (3.5) 25.7 (3.6) 25.7 (27.6, 23.7) ,0.001 21 21 Average RaO0–120 min (mmol z min z kg ) 23.8 (3.3) 23.9 (4.0) 0.1 (3.5) 24.9 (5.9) 7.6 (6.6) 217.3 (6.7) 216.9 (220.7, 213.1) ,0.001 2 2 Average EGP 0–240 min (mmol z min 1 z kg 1) 5.79 (1.76) 5.70 (2.48) 20.09 (1.86) 6.74 (2.10) 6.58 (1.21) 20.17 (2.45) 0.54 (20.91, 1.98) 0.452 2 2 Average EGP 0–120 min (mmol z min 1 z kg 1) 3.35 (1.93) 3.97 (2.35) 0.62 (2.41) 4.34 (2.18) 5.27 (1.38) 0.93 (2.44) 1.06 (20.42, 2.54) 0.152 Fasting insulin (mU z L21) 5.8 (3.4) 6.2 (4.1) 0.4 (1.4) 5.4 (4.0) 6.6 (4.2) 1.2 (1.5) 0.8 (20.3, 1.9) 0.137 Insulin iAUC 0–240 min (mU z L21 z min) 3,359 (2023) 3,822 (2,638) 463 (943) 5,365 (4,380) 3,068 (2,885) 21,791 (2,876) 21,301 (23,377, 775) 0.209 2 Insulin iAUC 0–120 min (mU z L 1 z min) 1,964 (1,329) 2,431 (2,052) 468 (879) 3,107 (2,377) 1,316 (1,575) 22,297 (4,997) 21,506 (22,868, 2143) 0.032 2 Fasting C-peptide (pmol z L 1) 752 (352) 744 (317) 28.0 (145) 678 (285) 711 (234) 33 (138) 25 (270, 119) 0.597 2 C-peptide iAUC 0–240 min (nmol z L 1 z min) 225 (74) 253 (134) 28 (94) 325 (166) 220 (164) 2104 (254) 239 (216, 84) 0.519 2 C-peptide iAUC 0–120 min (nmol z L 1 z min) 98 (41) 117 (85) 19 (55) 139 (66) 78 (62) 261 (103) 251 (2111, 9) 0.090 Fasting glucagon (pg z mL21) 66 (26) 62 (24) 24 (15) 63 (25) 66 (21) 3 (14) 7 (23, 16) 0.165 Glucagon iAUC 0–240 min (pg z mL21 z min) 648 (657) 426 (497) 2223 (612) 476 (616) 259 (369) 2317 (712) 2233 (2557, 91) 0.151 Glucagon iAUC 0–120 min (pg z mL21 z min) 581 (668) 350 (352) 2231 (585) 458 (618) 38 (71) 2420 (627) 2296 (2479, 2113) 0.003

HbA1c (%) 7.3 (0.6) 7.2 (0.6) 20.13 (0.29) 6.9 (0.4) 6.3 (0.5) 20.58 (0.25) 20.48 (20.70, 20.26) ,0.001 21 HbA1c (mmol z mol ) 56 (6.6) 55 (6.6) 21.4 (3.2) 52 (4.4) 45 (5.5) 26.3 (2.7) 25.2 (27.7, 22.8) ,0.001 2 2 Fasting insulin secretion rate (pmol z m 2 z min 1)* 93.2 (39.7) 93.7 (36.8) 0.5 (14.9) 84.5 (28.4) 102.8 (33.0) 18.3 (16.8) 17.0 (5.0, 29.0) 0.007 2 Total insulin secretion (nmol z m 2)* 53.5 (18.2) 57.7 (23.9) 4.2 (12.5) 64.8 (26.5) 59.2 (23.3) 25.5 (35.8) 22.3 (219.9, 15.2) 0.789 2 2 2 b-cell glucose sensitivity (pmol z min 1 z m 2 z mmol z L 1)* 20.2 (11.4) 22.8 (13.5) 2.6 (6.9) 29.4 (20.7) 53.7 (28.6) 24.3 (27.3) 2.10 (1.48, 2.98)^ ,0.001 Potentiation factor (ratio)* 1.0 (0.3) 1.1 (0.4) 20.02 (0.5) 1.0 (0.8) 1.2 (0.7) 0.1 (0.9) 0.4 (20.9, 0.1) 0.138 Rate sensitivity (pmol z mmol z L21 z m22)* 133 (219) 158 (373) 25 (342) 150 (237) 265 (1,209) 106 (1,105) 3.3 (0.0, 70,242)^ 0.811 Body weight (kg) 91.2 (17.1) 92.2 (17.0) 1.0 (1.6) 87.3 (14.2) 88.5 (14.4) 1.3 (2.0) 20.3 (21.6, 1.1) 0.714

Data are means (SD) unless otherwise indicated. EGP, rate of endogenous glucose production; RaO, Ra of oral glucose. *Data are median (IQR).^Indicates that the difference between groups is presented as the ratio of geometric means. care.diabetesjournals.org 6 Gastric Emptying With Sustained Lixisenatide Diabetes Care

major driver of gastrointestinal symp- toms induced by GLP-1 receptor agonists and is consistent with the recognition that, in general, there is a weak relation- ship between symptoms and delayed gastric emptying (32). The strengths of the current study are that optimal techniques were used to measure both gastric emptying and post- prandial glucose metabolism, in relation to the same glucose drink. The effects of lixisenatide on gastric emptying, sys- temic appearance of oral glucose, and postprandial glycemia were marked, and Figure 1—Gastric retention (%), measured by scintigraphy, of a 75-g glucose drink at baseline (day the dose and timing of administration of 0) and after 8 weeks’ treatment with placebo (n 5 15) or lixisenatide (n 5 15) (day 56) in lixisenatide conformed to standard clin- 30 metformin-treated patients with type 2 diabetes. Data are means (SD). The ratio of adjusted ical practice. One limitation of the study – geometric means for AUC (0 240 min) for lixisenatide and placebo at day 56 was 2.19 (95% CI 1.82, was that the patients in general had 2.64) (*P , 0.001) and for AUC (0–120 min) was 1.55 (95% CI 1.37, 1.74) (†P , 0.001). relatively good glycemic control and few complications, which may limit the gen- suitable for a short-acting GLP-1 receptor pancreas, lixisenatide directly stimulates eralizability of the observations. How- agonist. Conversely, patients with slow both first- and second-phase insulin re- ever, in more severely hyperglycemic gastric emptying at baselinedwhich is lease (28). Likewise, in patients with patients, achieving good control of fast- more common in those with multiple type 2 diabetes, lixisenatide enhances ing glycemia would be a higher priority complications of diabetes and poor over- both first- and second-phase insulin re- than would optimizing postprandial gly- all glycemic control (20)dwould be less sponse to an IV glucose challenge (29). In cemia, which was the target of interest likely to achieve substantial postprandial addition, chronically lower fasting and in the current study. Another limitation glucose lowering. postprandial glycemia reduces the im- was that the study was of only 8 weeks’ As well as slowing gastric emptying, pact of hyperglycemia/glucose toxicity duration and was not powered to ex- lixisenatide suppressed glucagon secre- onb-cellfunction.Ofnoteisthatbaseline amine relationships between changes in tion after the glucose drink, which rep- b-cell glucose sensitivity (;20 pmol z gastric emptying and HbA1c. Ideally, the resents an additional mechanism of min21 z m22 z mmol z L21) was already findings should be confirmed with a postprandial glycemic control. Likewise, profoundly impaired as compared with more physiological mixed solid and liq- postprandial plasma insulin concentra- the median value (105 pmol z min21 z uidmeal;nonetheless,a75-goralglu- tions were lower with lixisenatide, espe- m22 z mmol z L21) in a historical group cose load is a well-accepted standard for cially during the initial 2 h postprandially. of subjects without diabetes (n 5 96, evaluation of glucose tolerance, and Although endogenous glucose produc- age 5 55 years, BMI 5 30.3) receiving probably helped reduce heterogeneity tion is suppressed by GLP-1 in fasting 75 g oral glucose (23). Thus, lixisenatide in this proof-of-concept study. Further- pancreatic clamp experiments (26), its remains effective even in patients with more, we compared lixisenatide with postprandial regulation is more complex b-cell dysfunction (17). Therefore, the placebo rather than an active compar- (27). Concurrent suppression of both improved glycemic control induced by ator; it would now be of interest to glucagon and insulin, maintaining a rel- lixisenatide can be ascribed to both an examine effectson gastric emptying after atively constant ratio between them, indirect mechanism (slower gastric emp- prolonged exposure to a long-acting GLP- may explain why postprandial endoge- tying) and a direct action on the b-cell 1 receptor agonist. Also, we measured nous glucose production was similar with (30), although the multiple regression the effects of lixisenatide on gastric placebo and lixisenatide at day 56. Ad- analysis indicates that the effect on emptying after 8 weeks’ treatment but ditionally, the lack of change in endog- gastric emptying was predominant. not acutely, so we cannot be certain that enous glucose production despite lower Despite inducing profound slowing of there was not a slight decline in its effect plasma insulin concentrations could in- gastric emptying, lixisenatide was well over time. However, in our recent study dicate that hepatic insulin sensitivity was tolerated, and compliance was excellent. evaluating 15 patients with type 2 di- enhanced by lixisenatide. The presence of gastrointestinal adverse abetes by scintigraphy after a single Lixisenatide treatment had marked effects was specifically sought using a 10-mg dose of lixisenatide and a 75-g effects on b-cell function. Fasting insulin standardized questionnaire, and while glucose drink (21), gastric emptying over secretion rates were slightly reduced, these did occur in the early stages of the first 120 min showed a ;50% re- very likely on account of the lower blood treatment, they were mild and transient, duction with lixisenatide compared with glucose levels. In contrast, b-cell glucose as observed in previous analyses of self- placebo (mean 0.75 vs. 1.57 kcal z min21). sensitivity was greatly improved by lix- reported adverse effects from trials in- A similar calculation in the current study isenatide as compared with placebo. volving GLP-1 receptor agonists (31). This indicates mean gastric emptying of 0.56 With regard to the mechanisms under- highlights the fact that slowing of gastric kcal z min21 after lixisenatide and 1.95 lying these effects, in the perfused rat emptying per se is unlikely to be the kcal z min21 after placebo, i.e., a 70% care.diabetesjournals.org Rayner and Associates 7

Figure 2—Blood glucose concentration (A), Ra of oral glucose (B), rate of endogenous glucose production (C), and plasma insulin (D), C-peptide (E), and glucagon (F) concentrations before and after a 75-g glucose drink consumed at t 5 0 min at baseline (day 0) and after 8 weeks’ treatment with placebo (n 5 15) or lixisenatide (n 5 15) (day 56) in 30 metformin-treated patients with type 2 diabetes. Data are means (SD). *Significant adjusted group differences at day 56 for AUC (0–240 min); †significant adjusted group differences at day 56 for AUC (0–120 min). reduction. Although the lixisenatide dose exposure. Finally, it is possible that sup- glycemia by reducing the rate of small was higher in the current study, it seems pression of small intestinal motility and intestinal glucose absorption, as we have unlikely that there could have been much transit by lixisenatide could have con- previously shown with exenatide (33), diminutionineffect comparedwithacute tributed to the reduction in postprandial but we did not evaluate small intestinal 8 Gastric Emptying With Sustained Lixisenatide Diabetes Care

analyzed the data and reviewed the manuscript. K.L.J. analyzed scintigraphic data and reviewed themanuscript.M.H.contributedtostudydesign and data interpretation and reviewed the man- uscript. E.F. contributed to study design and data interpretation andreviewed the manuscript. D.T. contributed to study design and data interpre- tation. S.F. contributed to sample analysis and data interpretation. A.M. analyzed and inter- preted the data and reviewed the manuscript. A.N. contributed to study design, analyzed and interpreted the data, and reviewed the manu- script. C.K.R. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Prior Presentation. Parts of this study were presented in abstract form at the 54th Annual Meeting of the European Association for the Study of Diabetes, Berlin, Germany, 1–5 October 2018.

References 1. UK ProspectiveDiabetesStudy(UKPDS)Group. Intensive blood-glucose control with sulphonylur- Figure 3—Relationship between the change in postprandial blood glucose iAUC after 8 weeks’ eas or insulin compared with conventional treat- treatment with lixisenatide and the change in gastric retention AUC for 0–120 min in 15 patients ment and risk of complications in patients with treated with lixisenatide (Pearson r 520.89, †P , 0.001). type2diabetes (UKPDS33).Lancet1998;352:837– 853 2. Davies MJ, D’Alessio DA, Fradkin J, et al. function in the current study. However, supported by a University of Adelaide William Management of hyperglycaemia in type 2 di- the close relationship between the slow- T. Southcott Research Fellowship. abetes, 2018. A consensus report by the ing of gastric emptying and reduction in Duality of Interest. This investigator-initiated American Diabetes Association (ADA) and the study was supported financially by Sanofi. C.K.R. European Association for the Study of Diabetes Ra of the ingested glucose tracer suggests has received research funding from AstraZeneca, (EASD). Diabetologia 2018;61:2461–2498 that any contribution of small intestinal Merck Sharp & Dohme, Eli Lilly, and Company, 3. Kemp TM, Barr EL, Zimmet PZ, et al. Glucose, effects was probably modest. Novartis, and Sanofi and has participated in lipid, and blood pressure control in Australian Insummary,wehavedemonstrated, advisory boards for Allergan. T.W. has received adults with type 2 diabetes: the 1999-2000 travel support from Novartis and Sanofi and AusDiab. Diabetes Care 2005;28:1490–1492 using optimal techniques, that lixisena- research funding from AstraZeneca. K.L.J. has 4. Monnier L, Lapinski H, Colette C. Contribu- tide has profound effects to slow gastric received research funding from Sanofi and tions of fasting and postprandial plasma glucose emptying and retard the systemic ap- AstraZeneca and drug supplies from Merck Sharp increments to the overall diurnal hyperglycemia pearance of orally ingested carbohy- & Dohme. M.H. has participated in advisory of type 2 diabetic patients: variations with in- fi drate that persist with sustained use boards and/or symposia for Novo Nordisk, Sano , creasing levels of HbA(1c). Diabetes Care 2003; Novartis, Eli Lilly and Company, Merck Sharp & 26:881–885 and relate closely to the ability of this Dohme, Boehringer Ingelheim, and AstraZeneca 5. Riddle M, Umpierrez G, DiGenio A, Zhou R, medication to achieve marked lowering and has received honoraria for this activity. E.F. Rosenstock J. Contributions of basal and post- of postprandial glycemia. Short-acting has participated in advisory boards for Boeh- prandial hyperglycemia over a wide range of A1C GLP-1 receptor agonists are likely to be ringer Ingelheim, Eli Lilly and Company, and levels before and after treatment intensification fi – particularly applicable to patients with Sano ; has received research funding from Boeh- in type 2 diabetes. Diabetes Care 2011;34:2508 ringer Ingelheim and AstraZeneca; and has re- 2514 relatively good overall glycemic control ceived honoraria for speaking engagementsfrom 6. Deane AM, Nguyen NQ, Stevens JE, et al. andrapidoratleastnormalratesofgastric AstraZeneca, Sanofi, Boehringer Ingelheim, Eli Endogenous glucagon-like peptide-1 slows gas- emptying at baseline, irrespective of du- Lilly and Company, and Merck Sharp & Dohme. tric emptying in healthy subjects, attenuating ration of diabetes. Potentially, such pa- A.M. has received financial support from Boeh- postprandial glycemia. J Clin Endocrinol Metab – tientscouldbeidentifiedusingascreening ringer Ingelheim and is a consultant for Eli Lilly. 2010;95:215 221 A.N. has received financial support for clinical 7. Little TJ, Pilichiewicz AN, Russo A, et al. Effects test for gastric emptying, such as a stable research from Boehringer Ingelheim and Eli Lilly of intravenous glucagon-like peptide-1 on gastric isotope breath test (34), as a means of and Company. No other potential conflicts of emptying and intragastric distribution in healthy individualizing the choice of glucose- interest relevant to this article were reported. subjects: relationships with postprandial glyce- lowering medications, particularly since Sanofi had no involvement in the design or mic and insulinemic responses. J Clin Endocrinol development of the study or any role in the Metab 2006;91:1916–1923 we have shown that the rate of gastric collection, analysis, or interpretation of data. 8. Nauck MA, Niedereichholz U, Ettler R, et al. emptying remains relatively stable within Author Contributions. C.K.R. designed the Glucagon-like peptide 1 inhibition of gastric emp- individual patients when followed over a study, analyzed and interpreted the data, and tyingoutweighsitsinsulinotropiceffectsinhealthy number of years (35). drafted the manuscript. L.E.W. assisted with humans. Am J Physiol 1997;273:E981–E988 study design, recruited patients, and collected 9. Jones KL, Horowitz M, Wishart MJ, Maddox and analyzed data. L.K.P. recruited patients and AF, Harding PE, Chatterton BE. Relationships collected and analyzed data. K.L. undertook the between gastric emptying, intragastric meal Funding. Sanofi supplied lixisenatide and statistical analysis. M.J.B. recruited patients and distribution and blood glucose concentrations matching placebo. T.W. is supported by a Royal collected and analyzed data. J.G. recruited pa- in diabetes mellitus. J Nucl Med 1995;36:2220– Adelaide Hospital Florey Fellowship. K.L.J. is tients and collected and analyzed data. T.W. 2228 care.diabetesjournals.org Rayner and Associates 9

10. Gonlachanvit S, Hsu CW, Boden GH, et al. sitagliptin in patients with type 2 diabetes. J Clin morbidly obese patients. Diabetes Obes Metab Effect of altering gastric emptying on postpran- Endocrinol Metab 2012;97:2818–2826 2017;19:412–420 dial plasma glucose concentrations following a 19. Ewing DJ, Clarke BF. Diagnosis and manage- 28. Riz M, Pedersen MG, Toffolo GM, et al. physiologic meal in type-II diabetic patients. Dig ment of diabetic autonomic neuropathy. Br Med Minimal modeling of insulin secretion in the Dis Sci 2003;48:488–497 J (Clin Res Ed) 1982;285:916–918 perfused rat pancreas: a drug effect case study. 11. Nauck MA, Kemmeries G, Holst JJ, Meier JJ. 20. HorowitzM, Maddox AF,Wishart JM, Harding Am J Physiol Endocrinol Metab 2014;306:E627– Rapid tachyphylaxis of the glucagon-like peptide PE, Chatterton BE, Shearman DJ. Relationships E634 1-induced deceleration of gastric emptying in between oesophageal transit and solid and liquid 29. Meier JJ, Schenker N, Kahle M, Schliess F, humans. Diabetes 2011;60:1561–1565 gastric emptying in diabetes mellitus. Eur J Nucl Kapitza C, Menge BA. Impact of insulin glargine 12. Umapathysivam MM,Lee MY, Jones KL,et al. Med 1991;18:229–234 and lixisenatide on b-cell function in patients Comparative effects of prolonged and intermit- 21. Jones KL, Rigda RS, Buttfield MDM, et al. with type 2 diabetes mellitus: a randomized tent stimulation of the glucagon-like peptide Effects of lixisenatide on postprandial blood open-label study. Diabetes Obes Metab 2017; 1 receptor on gastric emptying and glycemia. pressure, gastric emptying and glycaemia in 19:1625–1629 Diabetes 2014;63:785–790 healthy people and people with type 2 diabetes. 30. Meier JJ, Kemmeries G, Holst JJ, Nauck MA. 13. Kapitza C, Forst T, Coester HV, Poitiers F, Ruus Diabetes Obes Metab 2019;21:1158–1167 Erythromycin antagonizes the deceleration of P, Hincelin-Mery´ A. Pharmacodynamic character- 22. Mari A, Stojanovska L, Proietto J, Thorburn gastric emptying by glucagon-like peptide 1 and istics of lixisenatide once daily versus liraglutide AW. A circulatory model for calculating non- unmasks its insulinotropic effect in healthy sub- once daily in patients with type 2 diabetes in- steady-state glucose fluxes. Validation and com- jects. Diabetes 2005;54:2212–2218 sufficiently controlled on metformin. Diabetes parison with compartmental models. Comput 31. Horowitz M, Aroda VR, Han J, Hardy E, Obes Metab 2013;15:642–649 Methods Programs Biomed 2003;71:269–281 Rayner CK. Upper and/or lower gastrointestinal 14. Drucker DJ, Buse JB, Taylor K, et al.; DURA- 23. Chang J, Russo A, Bound M, Rayner CK, Jones adverse events with glucagon-like peptide-1 re- TION-1 Study Group. Exenatide once weekly KL, Horowitz M. A 25-year longitudinal evalua- ceptor agonists: incidence and consequences. versus twice daily for the treatment of type 2 tion of gastric emptying in diabetes. Diabetes Diabetes Obes Metab 2017;19:672–681 diabetes: a randomised, open-label, non-inferi- Care 2012;35:2594–2596 32. Jones KL, Russo A, Stevens JE, Wishart JM, ority study. Lancet 2008;372:1240–1250 24. Matthews DR, Hosker JP, Rudenski AS, Berry MK, Horowitz M. Predictors of delayed 15. MeierJJ,RosenstockJ,Hincelin-MeryA,etal.´ Naylor BA, Treacher DF, Turner RC. Homeostasis gastricemptyingindiabetes.DiabetesCare2001; Contrasting effects of lixisenatide and liraglutide model assessment: insulin resistance and beta- 24:1264–1269 on postprandial glycemic control, gastric emp- cell function from fasting plasma glucose and 33. Thazhath SS, Marathe CS, Wu T, et al. The tying, and safety parameters in patients with insulin concentrations in man. Diabetologia glucagon-like peptide 1 receptor agonist exena- type2diabetesonoptimizedinsulinglarginewith 1985;28:412–419 tide inhibits small intestinal motility, flow, tran- or without metformin: a randomized, open-label 25. Watson LE, Xie C, Wang X, et al. Gastric sit, and absorption of glucose in healthy subjects trial. Diabetes Care 2015;38:1263–1273 emptying inpatientswithwell-controlledtype 2 and patients with type 2 diabetes: a randomized 16. Marselli L, Suleiman M, Masini M, et al. Are diabetes compared with young and older con- controlled trial. Diabetes 2016;65:269–275 we overestimating the loss of beta cells in type 2 trol subjects without diabetes. J Clin Endocrinol 34. Szarka LA, Camilleri M, Vella A, et al. A stable diabetes? Diabetologia 2014;57:362–365 Metab 2019;104:3311–3319 isotope breath test with a standard meal for 17. Bonadonna RC, Blonde L, Antsiferov M, et al. 26. Seghieri M, Rebelos E, Gastaldelli A, et al. abnormal gastric emptying of solids in the clinic Lixisenatide as add-on treatment among patients Direct effect of GLP-1 infusion on endogenous andinresearch.ClinGastroenterolHepatol2008; with different b-cell function levels as assessed glucose production in humans. Diabetologia 6:635–643.e1 byHOMA-bindex. DiabetesMetabRes Rev 2017; 2013;56:156–161 35. Watson LE, Phillips LK, Wu T, et al. Longi- 33:e2897 27. Camastra S, Astiarraga B, Tura A, et al. Effect tudinal evaluation of gastric emptying in type 2 18. Muscelli E, Casolaro A, Gastaldelli A, et al. of exenatide on postprandial glucose fluxes, diabetes. Diabetes Res Clin Pract 2019;154:27– Mechanisms for the antihyperglycemic effect of lipolysis, and ß-cell function in non-diabetic, 34