Economy & Finance

Hypoglycaemia with insulin aspart: a double-blind, randomised, crossover trial in subjects with Type 1 diabetes

To compare the effects of the rapid-acting insulin analogue insulin aspart and soluble human insulin on hypoglycaemia and glycaemic control in patients with Type 1 diabetes when injected immediately before meals as part of intensive insulin therapy.
of 7
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
  © 2004 Diabetes UK. Diabetic Medicine , 21 , 769–775  769 DOI: 10.1111/j.1464-5491.2004.01244.x Abstract Aims To compare the effects of the rapid-acting insulin analogue insulin aspartand soluble human insulin on hypoglycaemia and glycaemic control in patientswith Type 1 diabetes when injected immediately before meals as part of inten-sive insulin therapy. Methods In this multinational, double-blind, randomised, crossover trial, 155patients with Type 1 diabetes (HbA 1c  < 8.0%) were symmetrically randomisedto two 16-week treatment periods on either type of insulin, both injected 0–5 min before meals. NPH insulin was given as basal insulin once or twice dailyas needed, and insulin dosages were regularly adjusted using pre-defined algo-rithms to maintain tight glycaemic control. Treatment periods were separatedby a 4-week washout. Results The rate of major nocturnal (24.00–06.00 h) hypoglycaemic episodeswas 72% lower with insulin aspart than with human insulin (0.067 vs. 0.225events/month; P  = 0.001). Total rate of major hypoglycaemia did not differsignificantly between treatments (insulin aspart/human insulin relative risk0.72; 95% CI 0.47–1.09, P  = 0.12). The rate of minor events was significantlyreduced by 7% with insulin aspart ( P =  0.048). Reductions in rate of hypogly-caemia were achieved with maintained overall glycaemic control: Mean HbA 1c remained constant, slightly below 7.7% on both treatments. Conclusions The use of insulin aspart in an intensive insulin regimen inpatients with tightly controlled Type 1 diabetes led to clinically significantreductions in major nocturnal hypoglycaemia with no deterioration in gly-caemic control. Major nocturnal hypoglycaemia appears to be a strong clinicalindication for the use of rapid-acting insulin analogues during intensive insulintherapy.Diabet. Med. 21, 769–775 (2004) Keywords Type 1 diabetes, insulin aspart, hypoglycaemia, double-blind,clinical trial Abbreviations HI, human insulin; Iasp, insulin aspart; NS, not significant; SD ,standard deviation Correspondence to : Dr Simon Heller, Northern General Hospital, Sheffield S5 7AU, UK. E-mail: BlackwellPublishing,Ltd.Oxford,UKDMEDiabeticMedicine0742-3071BlackwellPublishing,200421OriginalArticleOriginalarticleReducedhypoglycaemiawithinsulinaspartS.R.Helleretal. Hypoglycaemia with insulin aspart: a double-blind, randomised, crossover trial in subjects with Type 1 diabetes S. R. Heller * , S. Colagiuri † , S. Vaaler ‡ , B. H. R. Wolffenbuttel § , K. Koelendorf ¶ , H. H. Friberg ** , K. Windfeld **  and A. Lindholm †† *Northern General Hospital, Sheffield, UK, †Prince of Wales Hospital, Randwick, Australia, ‡Solli Klinikk, Jessheim, Norway, §Department of Endocrinology, University Hospital Groningen, the Netherlands, ¶Diabetes out-patient clinic, Roskilde county hospital, Køge, Denmark, **Novo Nordisk A/S, Bagsværd, Denmark and ††Novo Nordisk Ltd, Crawley, UK  Accepted 12 November 2003  © 2004 Diabetes UK. Diabetic Medicine , 21 , 769–775 770 Reduced hypoglycaemia with insulin aspart • S. R. Heller et al. Introduction Rapid-acting insulin analogues have now been available forover 5 years. They exhibit more physiological pharmacoki-netic profiles compared with human or animal soluble insulinwith a faster rise to higher maximum insulin concentrationsand a more rapid fall to baseline [1,2]. The accompanyingimprovements in glucose profiles would be expected to pro-duce substantial clinical benefits [3]. However, the clinicaltrials of both available rapid-acting insulin analogues haveshowed relatively modest advantages over human insulin inmost studies. Results of various trials with either product haveconfirmed that their use in a basal bolus regimen reduces therisk of major hypoglycaemia by around 20% [2,4,5]. HbA 1c has fallen significantly in some studies [2,5–7], although itremains to be shown if these minor improvements of HbA 1c along with the more substantial reduction of post-prandialglucose peaks will reduce late complications in diabetes [3].In those trials where investigators have been able to per-suade patients to achieve particularly tight glycaemic targetsthe benefits appear to be greater, with more substantial andclinically useful falls in rates of major and nocturnal hypo-glycaemia [8,9]. However, these trials have utilized an opendesign based on the recommendation that soluble insulin shouldbe injected about 30 min before meals, whereas insulin ana-logues should be injected closer to mealtime. Many patientswith diabetes find these recommendations restrictive andgenerally inject human insulin immediately before meals, partlybecause this is often perceived as more convenient or practical,partly because of fear of hypoglycaemia before meals [10,11].Furthermore, the practice of injecting 20–40 min beforehandhas only minor benefits in reducing post-prandial glucose. Thepotential bias due to open designs in which patients believethey are taking more effective insulin raises questions as to thereliability of the conclusions of previous studies.We therefore set out to compare the effectiveness of therapid-acting analogue insulin aspart (IAsp) with that of solublehuman insulin (HI) in the setting of a basal bolus regimen usinga double-blind crossover design. Tight glycaemic control wasmaintained in order to investigate whether those at greatestrisk of hypoglycaemia were those who benefited the mostfrom the use of insulin analogues. Doses were adjusted using apre-defined algorithm to avoid bias caused by clinicians’greater experience with dosing human insulin. Patients and methods A double-blind, randomised, crossover trial was conductedin 19 sites in the UK, Denmark, Norway, Australia and theNetherlands. The trial was conducted after approval by localethics committees and was performed in accordance with theDeclaration of Helsinki. Written informed consent was obtainedfrom each patient before inclusion. A total of 155 patientswere randomised (by calls made to a central telephone number,using computer generated random numbers), of which 139completed the trial (Table 1). The two chief reasons for with-drawal during the trial were six patients who considered theirtreatment ineffective and four who failed to comply with thestudy protocol. Patients Eligible patients were males and females diagnosed withType 1 diabetes for at least 2 years, aged 18–65 years, with aBMI ≤  35 kg/m 2  and HbA 1c   ≤  9.0%. In total, 155 patients[35.7 ±  9.4 years (mean ±  standard deviation; SD ), BMI 24.0 ± 2.6 kg/m 2 , HbA 1c  8.6 ±  1.1%] were randomised. For the last3 months before the trial, patients were required to use humaninsulin before breakfast, lunch and dinner, and NPH insulin asrequired either once daily at bedtime, or twice daily beforebreakfast and at bedtime. Key exclusion criteria were impairedrenal or hepatic function, uncontrolled hypertension, cardiacproblems, presence of progressed late-diabetic complications,drug or alcohol abuse, and concurrent treatment with systemiccorticosteroids. Trial design Following a 6-week run-in, during which glycaemic controlwas tightened using a basal bolus regimen including solublehuman insulin, patients were symmetrically randomised todouble-blind treatment with insulin aspart (NovoRapid, NovoNordisk A/S, Bagsvaerd, Denmark) or human insulin (Actrapid,Novo Nordisk), both injected 0–5 min before meals. A cross-over design was used, such that each patient received 16 weeksof treatment with both preparations, in random order. NPHinsulin (Insulatard, Novo Nordisk) was given as basal insulin at   Treatment sequenceInsulin aspart followed by human insulinHuman insulin followed by insulin aspartCombinedRandomised8076156Withdrawals pre-exposure101Safety population7976155Withdrawals in period one448Withdrawals during wash-out202Withdrawals in period two336Efficacy population7370143Completed7069139 Table 1 Subject flow, randomization, withdrawal, completion and numbers included for statistical analyses  Original article  771 © 2004 Diabetes UK. Diabetic Medicine , 21 , 769–775 bedtime (and additionally before breakfast according to thepractice at the individual centre) throughout the trial. Injectionswere made using a pen injector (NovoPen 1.5, Novo Nordisk).Insulin doses were regularly adjusted to maintain tight gly-caemic control: seven-point blood glucose profiles, which wererecorded by the subjects weekly for 1 month and thereaftertwice monthly for 3 months, were used as input for a pre-defineddose adjustment algorithm (Table 2). Treatment periods wereseparated by a 4-week washout, where all patients were treatedwith human insulin before meals. Efficacy and safety evaluations Symptomatic hypoglycaemic episodes were recorded in diariesby the patients throughout the trial. Episodes were classified asminor if patients were able to deal with the episode themselves.Hypoglycaemic episodes were classified as major if patientsrequired help from another person by administration of oralglucose, intravenous glucose or glucagon treatment.HbA 1c  was measured at screening and at the beginning andend of each treatment period using the Bio-Rad-VARIANTion-exchange high performance liquid chromatography assay(Bio-Rad, Hercules, CA, USA; coefficient of variation approx.1.5%). Subjects also recorded nine-point blood glucose profileson one day during the last week of each treatment period.Measurements were performed using new and calibrated bloodglucose meters [in Europe the One Touch® II (LifeScan, Milpitas,CA, USA); in Australia the Medisense (Abbott, Abbott Park, IL,USA)] at the following time points: before and 90 min aftermeals, at bedtime, at 02.00 h, and before breakfast the follow-ing day. Statistical analysis The primary endpoint was the rate of severe hypoglycaemia,while secondary endpoints included symptomatic hypogly-caemia, HbA 1c  and weight. The number of participants was basedon power calculations using hypoglycaemia rates observed ina previous and similar study [12]. We expected 400–500 majorhypoglycaemic events, which would have allowed detection of a relative difference of 30–40% in risk of major hypoglycaemiawith a probability of approximately 70–95%. Data analysiswas based on the efficacy population comprising all subjectswith data from both treatment periods.Relative risks of major and minor hypoglycaemia with thetwo treatments were estimated by logistic regression, correctingfor relative exposure time on the two treatments. Diurnaldistribution of major hypoglycaemic episodes was analysedsimilarly, separating daytime (06.00–24.00 h) and night-time(24.00–06.00 h): events. The P -values for treatment differenceswere based on the Wald test [13]. The effect of treatment onHbA 1c  as recorded at the end of treatment periods and on bloodglucose at selected time-points of the nine-point blood glucoseprofiles was analysed using a standard Armitage-Hill crossoveranalysis [14], adjusting for sequence and period effects. Data arepresented as mean ± SD  together with 95% CI where appropriate. P -values were based on t  -tests and the significance level setto 0.05. Time of blood glucose testResult (m M  blood glucose)Insulin adjustment*Pre-breakfast< 5.0Reduce bedtime NPH by 2 U5.0–7.0No change> 7.0Increase bedtime NPH by 2 U90 min post-breakfast< 5.0Reduce pre-breakfast bolus insulin by 2 U5.0–9.0No change> 9.0Increase pre-breakfast bolus insulin by 2 UPre-lunch (no NPH < 5.0Reduce pre-breakfast bolus insulin by 2 Uprebreakfast)5.0–7.0No change> 7.0Increase pre-breakfast bolus insulin by 2 UPre-lunch (with < 5.0Reduce pre-breakfast NPH by 2 UNPH pre-breakfast)5.0–7.0No change> 7.0Increase pre-breakfast NPH by 2 U90 min post-lunch< 5.0Reduce pre-lunch bolus insulin by 2 U5.0–9.0No change> 9.0Increase pre-lunch bolus insulin by 2 UPre-dinner< 5.0Reduce pre-lunch bolus insulin by 2 U5.0–7.0No change> 7.0Increase pre-lunch bolus insulin by 2 U90 min post-dinner< 5.0Reduce pre-dinner bolus insulin by 2 U5.0–9.0No change> 9.0Increase pre-dinner bolus insulin by 2 UBedtime< 5.0Reduce pre-dinner bolus insulin by 2 U5.0–7.0No change> 7.0Increase pre-dinner bolus insulin by 2 U*The dose adjustment was 2 U unless: (i) the individual dose was = 20 U, in which case an adjustment of 4 U was suggested; (ii) a clinical need for a larger adjustment was indicated; (iii) the patient wished more cautious adjustment in steps. Table 2 Algorithm for adjustment of insulin dose  © 2004 Diabetes UK. Diabetic Medicine , 21 , 769–775 772 Reduced hypoglycaemia with insulin aspart • S. R. Heller et al. The statistical analysis was conducted using SAS v.6.12 on aUNIX platform. Results Insulin doses The average daily dose of bolus insulin (expressed as IU ×  kg − 1 ×  day − 1 ; mean ±   SD ) was constant for the two preparationsthroughout treatment periods, with 0.46 ±  0.17 for insulinaspart and 0.47 ±  0.18 for human insulin at the end of treat-ment periods. Likewise, average daily dose of basal insulin wassimilar for the two treatments, with doses at the end of treat-ment periods of 0.31 ±  0.12 for insulin aspart and 0.29 ±  0.11for human insulin groups. Also, number and timing of NPHinjections were similar between treatments: at the beginning of treatment periods, 15% of patients were on a regimen of basalinsulin injection both at breakfast and bedtime. At the end of treatment periods this proportion had increased to 25% withinsulin aspart and 20% with human insulin. Glycaemic control HbA 1c  was 7.9 ±  0.7% (mean ±   SD ) at baseline and fell to 7.7 ± 0.8% by the end of the run-in period. At the end of thetreatment periods, the levels of HbA 1c  did not differ betweentreatments, being 7.7 ±  0.8% for insulin aspart and 7.7 ± 0.9% for human insulin (difference 95% CI: − 0.105–0.170; P  = 0.64).The nine-point blood glucose profiles (Fig. 1) showed signific-antly lower post-prandial glucose concentrations with insulinaspart 90 min after breakfast ( P =  0.0001) and 90 min afterdinner ( P =  0.023). Blood glucose concentrations were signific-antly higher with insulin aspart before dinner ( P =  0.0001), at02.00 h ( P =  0.0015) and before breakfast on the followingmorning ( P =  0.030). Hypoglycaemia Results on hypoglycaemia are summarized in Table 3. Therewere 38 major hypoglycaemic episodes during 44.5 subjectyears of treatment with insulin aspart (0.85 events per patient-year), and 51 episodes during 45.8 subject years of treatmentwith human insulin (1.11 events per patient-year). The result-ing risk ratio of 0.72 was not significantly different from unity(95% CI: 0.47–1.09; P  = 0.12). However, major episodesoccurring during the night (24.00–06.00 h) were significantlyreduced with insulin aspart (risk ratio 0.28 95% CI: 0.13–0.59; Figure 1 Self-monitored blood glucose profiles obtained during the last week of treatment periods. Data are means ±  2 standard error of the mean. An asterisk denotes a significantly different value compared with the other pre-meal insulin; human insulin (  , solid line), insulin aspart (  , dashed line). Table 3 Hypoglycaemic events during treatment with insulin aspart and human insulin   Insulin aspart EventsHuman insulin Events n Pt-year − 1 n Pt-year − 1 RR (95% CI)* P -value† ExposedMinor159035.8175238.20.93 (0.87–1.00)0.048Major all380.85511.120.72 (0.47–1.09)0.12Major night90.80312.700.28 (0.13–0.59)0.001Major daytime290.86200.581.38 (0.78–2.45)0.27Grade A all‡280.64400.880.67 (0.41–1.08)0.10Grade A night60.54242.10.24 (0.10–0.59)0.002Grade A daytime220.66160.471.30 (0.68–2.48)0.42Grade B all§100.23110.240.89 (0.38–2.10)0.79Grade B night30.2670.610.42 (0.11–1.63)0.21Grade B daytime70.2040.121.71 (0.50–5.88)0.39*The estimated relative risk is based on a statistical model conditioning on the total number of events per subject and correcting for relative exposure time on IAsp and HI.†The P -values are based on the Wald test.‡Grade A: the patients received oral glucose.§Grade B: the patients required intravenous glucose or glucagon treatment.  Original article  773 © 2004 Diabetes UK. Diabetic Medicine , 21 , 769–775 P  = 0.001). There were nine major nocturnal episodes withinsulin aspart, equivalent to about four major nocturnalepisodes per five patient-years. There was no significant differ-ence in the daytime risk of major hypoglycaemia, the estimatedIAsp/HI relative risk being 1.38 (95% CI: 0.78–2.45; P  = 0.27).Minor hypoglycaemic episodes were reduced with insulinaspart compared with human insulin (1590 events vs. 1752events, the estimated IAsp/HI relative risk being 0.93 (95% CI:0.87–0.999; P  = 0.048).Of those patients who used twice daily basal insulin through-out, two patients experienced a total of five major episodeswhen on insulin aspart and three experienced a total of sevenmajor episodes using human soluble insulin. Discussion In this double-blind study, the substitution of the rapid-actinginsulin analogue insulin aspart as part of a basal bolus insulinregimen in tightly controlled patients reduced the risk of majornocturnal hypoglycaemia. Minor hypoglycaemic episodeswere also reduced significantly, although by a relativelymodest amount. There was no significant difference in thetotal number of major hypoglycaemic episodes, but there wasa trend towards a lower rate when patients were taking insulinaspart. These findings are in agreement with previous large-scaleopen-labelled studies of insulin aspart [5,15] and some studiesof insulin lispro [4,6,16,17]. The apparent failure to demon-strate a difference in the overall frequency may be due to atype 2 statistical error. We calculated the power of our studybased on a previous double-blind study of tightly controlledpatients using insulin aspart [12]. However, the rates of majorepisodes in the current study were lower than anticipated andmore comparable with larger non-blinded studies [2,5].Our most clinically relevant finding was a significant fallin major nocturnal hypoglycaemic episodes of around 70%when participants used insulin aspart and this was achieved atthe cost of a slight increase in nocturnal blood glucose levelsrelative to those obtained with human insulin. A similar patternwas observed in a double-blind comparison of insulin lisproand human insulin with a similar study design [18], althoughin this study there was no significant reduction in majornocturnal episodes with lispro. Overall frequency of symptomatichypoglycaemia did not differ between treatments, but patientshad a significantly reduced risk of hypoglycaemia between24.00 and 06.00 h with insulin lispro, coinciding with a higherblood glucose level during the night.The raised nocturnal blood glucose levels with insulin aspartoccurred despite similar evening doses of NPH insulin. Thisfinding confirms that insulin aspart contributes less to theinsulin supply during the night than does human insulin. Slightlyhigher nighttime blood glucose levels should benefit patientsat high risk of nocturnal hypoglycaemia. For patients at lowerrisk, the relative hyperglycaemia during night-time withinsulin aspart might allow for a higher evening dose of NPHinsulin, thereby improving night-time and pre-breakfastglycaemic control. Indeed, it is disappointing that the pre-defined dosing algorithm did not result in better fastingglycaemic control with insulin aspart than was the case. Possibly,the seven-point blood glucose profiles, which were recordedevery 1–2 weeks, may not have provided enough informationto adjust the evening NPH dose efficiently. Blood glucoseprofiles may need to be measured more frequently to enablemore effective insulin dose adjustment and, in the future,more detailed information using continuous glucose monitor-ing may lead to better glycaemic control without increasinghypoglycaemia [19].The reduction in rates of severe nocturnal hypoglycaemiawhen subjects used insulin aspart were not matched by areduced risk in daytime episodes, indeed there was a trend tomore episodes during the day. The reasons for this remainunclear and, as the precise timing of daytime episodes was notrecorded, we were unable to determine whether these episodesclustered at particular times of the day. However, it is of inter-est that in the double-blind study of insulin lispro, an increasedfrequency of late-morning hypoglycaemia was seen in patientstaking insulin lispro compared with conventional solubleinsulin [18]. Perhaps the lower post-prandial glucose valuesobserved with rapid acting insulin analogues after breakfast,increase the chances of hypoglycaemia before the middaymeal, particularly if a mid-morning snack is omitted.In this study, treatment with insulin aspart tended tolower post-prandial glucose, but overall glycaemic control, asassessed by HbA 1c , was not improved when compared withhuman insulin. Pre-prandial blood glucose was about equalbefore breakfast and lunch, but significantly higher withinsulin aspart before dinner. This was probably due to theshorter duration of action of insulin aspart, combined with anattenuation of the effect of NPH insulin in the late afternoon,as most patients injected NPH insulin at bedtime only. Withhuman insulin, the longer duration of action partly compens-ated for the lack of basal insulin before dinner.Rapid-acting insulin analogues demonstrate more physio-logical insulin profiles than standard human soluble insulinand it was hoped that their use in multiple insulin regimensmight improve glycaemic control and lead to less hypoglycae-mia. However, experience with rapid-acting insulin analoguesshows that it is difficult to improve both outcomes at the sametime. Glycaemic control is usually unaltered, although somestudies have shown significant, albeit minor, falls in HbA 1c  of around 0.3% or less [5–7]. Nevertheless, in those who attemptto maintain particularly tight glucose targets, there is evidencethat rapid-acting insulin analogues lead to clinically usefulreductions in hypoglycaemia when compared with solublehuman insulin [20]. Hollemann et al  . [9] used a combinationof lispro and isophane insulin in a group of patients whoattained tight glycaemic control with a mean HbA 1c  of approx-imately 7.3% and a 38% reduction in major hypoglycaemicepisodes on insulin lispro. Heller et al  . [8] also demonstrated a71% reduction in symptomatic nocturnal episodes in patientswho achieved particularly tight glycaemic control; HbA 1c
Similar documents
View more...
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks