Insulin Degludec, An Ultra-longacting Basal Insulin .

ArticlesSafety variables were adverse events, hypoglycaemicepisodes, insulin dose, bodyweight, injection-site reactions, abnormal findings related to physical examination,vital signs, fundoscopy, electrocardiogram (ECG), andlaboratory tests (including antibodies). Confirmed hypoglycaemic episodes included those with a plasma glucoseconcentration of less than 3·1 mmol/L or severe episodesnecessitating assistance3 (appendix p 7). Hypoglycaemicepisodes occurring from 0001 h and 0559 h were classifiedas nocturnal and those occurring from 0600 h and 0000 hwere classified as diurnal.Laboratory analyses were undertaken at the commercialcentral laboratories (Quintiles Laboratories Europe, WestLothian, UK, Quintiles Laboratories South Africa, Irene,South Africa, Quintiles Laboratories, Marietta, GA, USA).Antibodies were analysed at Celerion Switzerland,Fehraltorf, Switzerland, by a validated subtraction radioimmunoassay method.20,21722 participants assessed for eligibility93 excluded55 did not meet inclusion criteria*23 met exclusion criteria6 outside randomisation period5 withdrew informed consent3 unavailable1 lost to follow-up629 enrolled and randomly assigned 3:1472 allocated to insulin degludecAllocation157 allocated to insulin glargine3 (2%) withdrawn1 ( 1%) professional reason2 (1%) unwilling to adhere totreatment regimenStatistical analysisThe primary objective was to confirm the non-inferiorityof insulin degludec to insulin glargine in reduction inHbA1c from baseline after 52 weeks of treatment.19 Noninferiority was confirmed if the upper limit of the 95% CIof the treatment difference was less than or equal to0·4% points, as recommended by regulatory guidelines.The type 1 error rate was controlled by use of ahierarchical (fixed-sequence) testing procedure for selectedendpoints, including change in HbA1c concentration,number of nocturnal confirmed and overall confirmedhypoglycaemic episodes, and change in fasting plasmaglucose (FPG; appendix p 8). Sample size was determinedby the primary objective with the assumption of a onesided t test at a significance level of 2·5%, a zero meantreatment difference, and an SD of 1·1% for HbA1c. A totalof 624 participants were needed for at least 95% powerafter adjustment for a 15% dropout rate.All participants randomly assigned treatment wereincluded in the intention-to-treat statistical analyses of allefficacy endpoints (HbA1c, FPG, SMPG, health-relatedquality of life [HRQoL]), hypoglycaemia, bodyweight, andlipids. All other safety endpoints were assessed inparticipants exposed to treatment. Missing values wereimputed by the last observation carried forwardapproach.19 Statistical analyses were done with SASsoftware (version 9.1.3).Baseline characteristics, demography, and adverseevents were presented by use of descriptive statistics.Treatment differences in HbA1c, FPG, SMPG, HRQoL,insulin dose, bodyweight, and lipid concentrations after52 weeks of treatment were estimated by use of ANOVA,with treatment, antidiabetic treatment at screening (oncedaily or more than once-daily basal injections [includingpump]), sex, and region as fixed factors, and age andbaseline values as covariates. Rate ratios of hypoglycaemicepisodes were estimated by use of a negative binomialregression model with treatment, antidiabetic therapy atwww.thelancet.com Vol 379 April 21, 2012472 (100%) treatedFollow-up17 (11%) withdrawn2 (1%) adverse events†2 (1%) non-compliance3 (2%) withdrawal criteria¶10 (6%) other reasons 68 (14%) withdrawn12 (3%) adverse events†11 (2%) non-compliance2 ( 1%) ineffective therapy‡15 (3%) withdrawal criteria§28 (6%) other reasons 137 (87%) completed treatment404 (86%) completed treatment472 (100%) full-analysis set448 (95%) per-protocol analysis set472 (100%) safety-analysis set154 (98%) treatedAnalysis157 (100%) full-analysis set147 (94%) per-protocol analysis set154 (98%) safety-analysis setFigure 1: Trial profileFull-analysis set included all randomly assigned participants. Safety-analysis set included participants exposed totreatment. Per-protocol analysis set included participants who complied with all recruitment criteria (ie, inclusionand exclusion criteria), had at least 12 weeks of exposure, and had valid assessments of glycated haemoglobin atbaseline and at or after 12 weeks of treatment. *37 participants did not meet inclusion criteria of glycatedhaemoglobin of 10% (86 mmol/mol) or less. †See appendix p 9. ‡As judged by investigator. §Two participantswere withdrawn because of lack of effect as per withdrawal criterion. ¶One participant was withdrawn because oflack of effect as per withdrawal criterion. See appendix p 9.screening, sex, and region as fixed factors, and age ascovariate, for all reported episodes that were assessed tobe treatment-emergent in all randomly assignedparticipants (predefined analysis). To establish thehypoglycaemic profile after achievement of stable doseand glycaemic control for most participants, the modelwas also fitted in a post-hoc analysis of episodes occurringin the maintenance period from week 16 to week 52. The9-point SMPG profile data were analysed with a repeatedmeasures model. Prandial increments in plasma glucosewere analysed by use of the ANOVA method describedabove. The time to first achieve SMPG before breakfastof less than 5·0 mmol/L was analysed by use of a Coxproportional hazards model. Insulin dose and diurnalconfirmed hypoglycaemia were analysed post hoc. Data1491

ArticlesInsulin degludec Insulin glargineFull-analysis set472157Safety-analysis set472154Women194 (41%)67 (43%)White437 (93%)148 (94%)Black9 (2%)3 (2%)Asian6 (1%)3 (2%)Other20 (4%)3 (2%)RaceEthnic originHispanic or Latin American22 (5%)10 (6%)Non-Hispanic or non-Latino450 (95%)147 (94%)Age (years)42·8 (13·7)Bodyweight (kg)78·9 (14·3)78·3 (16·2)Body-mass index (kg/m²)26·3 (3·7)26·4 (4·2)Duration of diabetes (years)19·1 (12·2)18·2 (11·4)HbA1c (%)HbA1c (mmol/mol*)FPG (mmol/L)43·7 (13·3)7·7 (0·9)7·7 (1·0)60·7 (9·8)60·7 (11·0)9·1 (4·0)9·7 (4·4)Diastolic blood pressure (mm Hg)74·1 (8·4)73·5 (7·9)Systolic blood pressure (mm Hg)121·8 (12·8)119·2 (12·9)HDL cholesterol (mmol/L)1·57 (0·4)1·60 (0·5)LDL cholesterol (mmol/L)2·51 (0·8)2·47 (0·8)Total cholesterol (mmol/L)4·54 (0·9)4·55 (0·9)Triglycerides (mmol/L)1·00 (0·9)1·03 (0·7)Basal insulin type at screeningGlargine334 (71%)108 (69%)Detemir87 (18%)34 (22%)Neutral protamine Hagedorn37 (8%)12 (8%)Other2 ( 1%)··Bolus insulin at screeningAspart244 (52%)81 (52%)Lispro183 (39%)59 (38%)Other45 (10%)17 (11%)Data are number, number (%), or mean (SD). Data are for the full-analysis set,with the exception of lipids and blood pressure, which are for the safety-analysisset. Baseline values for FPG, HbA1c, and lipids were recorded at the randomisationvisit (week 0) and the others were recorded at the screening visit (1 week beforerandomisation). HbA1c glycated haemoglobin. FPG fasting plasma glucose.*Calculated, not measured, by the formula: (HbA1c [%] – 2·15) 10·929.Table 1: Demographics and baseline characteristics of the randomlyassigned population in the insulin degludec and insulin glargine groupswere reported with 95% CI and p values for two-sidedtesting at an α of 0·05.This trial is registered with ClinicalTrials.gov, numberNCT00982228.Role of the funding sourceThe sponsor was responsible for the study design, supplyof trial products and equipment, monitoring, datamanagement, statistics, and preparation of the clinicaltrial report. All authors had access to trial data and had fullresponsibility for the content of the report and had finalresponsibility for the decision to submit for publication.1492Results626 of 629 participants who were randomly assigned totreatment between Sept 1, 2009, and Nov 8, 2010, weregiven one of the trial drugs, and most (404 [86%] of 472 ininsulin degludec group and 137 [87%] of 157 in insulinglargine group) completed the trial (figure 1). The overallwithdrawal pattern was similar in the two groups.Baseline characteristics were representative of a population with type 1 diabetes mellitus with reasonably goodglycaemic control (mean HbA1c 7·7%, 60·7 mmol/mol,table 1). The antidiabetic regimen before the trial for mostparticipants (455 [72%] of 629) consisted of once-dailybasal injection with one or more bolus doses; insulinglargine and insulin aspart were the most commonlyused pretrial basal and bolus insulins, used by 442 (70%)and 325 (52%) participants, respectively (table 1).As would be expected with the treat-to-target method,the mean decrease in HbA1c from baseline was similarbetween treatments (figure 2A; appendix p 10): 0·40%points (SE 0·03) for insulin degludec and 0·39% points(0·07) for glargine with an estimated treatment difference(ETD) of –0·01% points (95% CI –0·14 to 0·11; p 0·0001for one-sided test of non-inferiority evaluated at the 2·5%level). Thus, insulin degludec was non-inferior to insulinglargine in reducing HbA1c concentrations. The robustness of the primary analysis was further supported by theresults of the per-protocol set (–0·01% points [–0·14 to0·12]) and additional sensitivity analyses (appendix p 11).Similar proportions of participants achieved the ADAand EASD HbA1c target ( 7%, 53 mmol/mol) with insulindegludec (188 [40%] of 472) and insulin glargine (67 [43%]of 157). After 52 weeks of treatment, mean baseline FPGdecreased by 1·3 mmol/L (SE 0·2) to 7·8 mmol/L (0·2)with insulin degludec and by 1·4 mmol/L (0·4) to8·3 mmol/L (0·3) with insulin glargine. The meanreduction in laboratory-reported FPG (figure 2B; appendix p 10) was not significantly different between treatments,with an estimated treatment difference of –0·33 mmol/L(95% CI –1·03 to 0·36; p 0·35). Mean 9-point SMPGprofiles decreased in both groups (figure 2C). The meanSMPG before breakfast used for dose adjustment decreasedfrom 8·6 mmol/L (0·1) at baseline to 7·3 mmol/L (0·1)with insulin degludec and from 8·6 mmol/L (0·2) to7·8 mmol/L (0·2) with insulin glargine. The mean SMPGbefore breakfast was significantly lower with insulindegludec; estimated treatment difference between insulindegludec and insulin glargine was –0·55 mmol/L(–1·03 to –0·08; p 0·023). Differences in the mean prandial increments after breakfast (p 0·70), lunch (p 0·68),and main evening meal (p 0·86) were not significant(appendix p 11). The median time to first achieve the SMPGbefore breakfast of less than 5·0 mmol/L was 5 weeks(IQR 3–14) for participants treated with insulin degludecand 10 weeks (4–22) for those given insulin glargine(appendix p 12).At end of trial, the mean values for daily basal, dailybolus, and daily total insulin doses were significantlywww.thelancet.com Vol 379 April 21, 2012

Articles64Insulin degludec (n 472)Insulin glargine (n 157)7·9627·8HbA1c 354452FPG 1447·51357·0126FPG (mg/dL)000481216202428Time (weeks)323640444852C12204Insulin degludec (n 472)Insulin glargine (n 157)119-point SMPG (mmol/L)216Week05219218010168915614489-point SMPG (mg/dL)www.thelancet.com Vol 379 April 21, 2012A8·0HbA1c m9090tfasakreebfort0time0r120st7Belower by 14%, 10%, and 11%, respectively, in the insulindegludec group relative to the insulin glargine group(table 2). Mean daily doses of insulin degludec changedlittle during the trial, and there was a substantial increasein insulin glargine in the first 5 weeks. The mean dailyinsulin aspart doses increased mostly during the first12 weeks in both treatment groups (appendix p 13).Mean weight gain was similar in both treatmentgroups: 1·8 kg (SE 0·2) with insulin degludec and 1·6 kg(0·3) with insulin glargine (p 0·62; appendix p 11, p 14).The rates of all confirmed hypoglycaemic episodes anddiurnal confirmed hypoglycaemic episodes were similarin the two treatment groups (figure 3A; table 3). Thelower rate of nocturnal confirmed hypoglycaemia withinsulin degludec was apparent as early as 8 weeks aftertreatment initiation, and was significantly lower by 25%at the end of treatment. The rate of severe hypoglycaemicepisodes was low and did not differ significantly betweentreatments (table 3). The rates of overall confirmed andnocturnal confirmed hypoglycaemic episodes were alsoanalysed in participants during the maintenance phase(from week 16 to end of trial) when insulin doses andglycaemic indicators seemed to have stabilised for mostparticipants. In the maintenance phase, the rate ofconfirmed hypoglycaemic episodes per patient-year ofexposure was similar between the insulin degludec andinsulin glargine groups (37·30 vs 36·22, estimated rateratio of insulin degludec to insulin glargine 1·02 [95% CI0·83–1·25]; p 0·83) and the rate of nocturnal confirmedhypoglycaemic episodes per patient-year of exposure wassignificantly lower by 27% with insulin degludec (3·91 vs5·22, 0·73 [0·56–0·96]; p 0·024).Adverse event rates were 438 per 100 patient-years ofexposure (PYE) for insulin degludec and 432 per 100 PYEfor insulin glargine (table 4). No treatment-specificpatterns were seen (appendix p 15). Most adverse events(2377 [94%] of 2518) were mild or moderate and werethought to be unrelated to basal insulin by the investigator(2315 [92%]; table 4; appendix p 30). The rate of injectionsite reactions was low in both groups (table 4) and nonewere severe.Rates of serious adverse events per 100 PYE were similarbetween the insulin degludec and insulin glargine groups.The distribution of the number of serious adverse eventswas similar in the two treatment groups (appendix p 25)and most participants in the insulin degludec (423 [90%] of472) and insulin glargine (137 [89%] of 154) groups reportedno serious adverse events. The most frequently reportedserious adverse events related to basal insulin werehypoglycaemia, hypoglycaemic unconsciousness, andhypoglycaemic seizure (appendix p 26). Four seriousadverse events were adjudicated as major adversecardiovascular events, of which three were fatal. Oneserious adverse event of sudden death was in a 26-year-oldwoman who was found dead in bed after 32 days oftreatment; her medical history included diabetes andasthma. The death was judged to be related to insulinFigure 2: Glycaemic efficacy in the insulin degludec and insulin glargine groups(A) HbA1c. (B) FPG. (C) 9-point SMPG at baseline and end of treatment. Data are mean (SE), reported for allrandomly assigned participants (full-analysis set). Error bars represent SE. Last observation carried forwardapproach was used for imputing postbaseline missing data. FPG fasting plasma glucose. HbA1c glycatedhaemoglobin. SMPG self-measured fasting plasma glucose.1493

ArticlesInsulin degludec Insulin glargine(U/kg; n 470)(U/kg; n 154)Mean ratio*of insulindegludec toinsulinglargineEstimatedtreatmentratio† of insulindegludec toinsulin glargine(95% CI)p valueBasal insulin (degludec or glargine)Pretrial0·37 (0·01)0·36 (0·01)····Week 10·35 (0·01)0·33 (0·01)······Week 520·35 (0·01)0·39 (0·02)0·910·86 (0·81–0·92)·· 0·0001Bolus insulin aspart (at mealtimes)Week 10·36 (0·01)0·38 (0·01)····Week 520·40 (0·01)0·44 (0·02)0·900·90 (0·83–0·98)··Total insulin (basal bolus)0·016··Week 10·72 (0·01)0·71 (0·02)····Week 520·75 (0·02)0·82 (0·03)0·910·89 (0·84–0·93) 0·0001Data are mean (SE), unless otherwise indicated, for all participants exposed to treatment who contributed to analyses.Missing values at week 52 were imputed with the last observation carried forward approach. *Calculated as the ratio ofmean doses of insulin degludec to insulin glargine at last treatment visit. †Estimated by use of ANOVA oflog-transformed dose value (U/kg) at week 52 with treatment, antidiabetic therapy at screening, sex, and region asfixed factors, and age and week 1 dose as covariates.Table 2: Daily insulin dose in the insulin degludec and insulin glargine groupsAOverall confirmed hypoglycaemia(cumulative events per participant)45Insulin degludec (n 472)Insulin glargine (n 154)4035302520151050BNocturnal confirmed hypoglycaemia(cumulative events per participant)6543210048121620 24 28 32Time (weeks)36 40 44 4852Figure 3: Overall confirmed (A) and nocturnal confirmed (B) hypoglycaemicepisodesData are for all exposed participants (safety-analysis set). Insulin exposureperiod was from first day of treatment to no later than 7 days after the last dayof treatment.glargine and insulin aspart by the investigator. Two fatalmyocardial infarctions were reported in the insulindegludec group and were not causally related to treatment—1494one occurred after 79 days of treatment in a 67-year-oldman whose medical history included diabetes,hypertension, and cardiovascular disease, and the otherdeath occurred after 138 days of treatment in a 60-year-oldman whose medical history included diabetes and obesity.The median concentration of antibodies specific toinsulin degludec was zero during the trial and theconcentration of antibodies cross-reacting betweeninsulin degludec and human insulin stayed low duringthe trial (appendix p 27). No apparent association betweenthe development of insulin-degludec-specific antibodiesor cross-reacting antibodies and HbA1c or insulin dosewas noted (data not shown).No differences were noted in physical examinationfindings, vital signs, electrocardiography, fundoscopy,and laboratory measurements. No significant differenceswere seen in HRQoL assessments with the 36-item shortform health (SF-36) survey (version 2; appendix p 28).DiscussionReduction in HbA1c concentration from baseline withinsulin degludec and insulin glargine was similar, thusestablishing non-inferiority of insulin degludec to insulinglargine in improving long-term glycaemic control intype 1 diabetes. Further evidence of improved glycaemiccontrol with both insulins was the reduction in FPG andSMPG. Mean SMPG before breakfast was significantlylower with insulin degludec than with insulin glargine;reduction in laboratory-reported FPG tended to be greaterwith insulin degludec than with insulin glargine, thoughthe difference was not significant.The before-breakfast titration target of 3·9 mmol/L toless than 5·0 mmol/L is lower than targets describedin other intensive basal-bolus therapy trials in patientswith type 1 diabetes—eg, preprandial SMPG target of3·9–6·7 mmol/L in the DCCT study.2 The flat and stablepharmacokinetic profile of insulin

Novo Nordisk, Bagsværd, Denmark) or insulin glargine (Lantus, 100 U/mL, subcutaneously, 3 mL SoloStar, insulin and insulin pen manufactured by Sanofi , Paris, France), both in combination with meal-time insulin aspart (NovoRapid/NovoLog, 100 U/mL, subcutaneously, 3 mL FlexPen, Novo Nordisk, Bagsvaerd, Denmark), by means