Efficacy and Safety of Rapid-Acting Insulin Analogs in Special Populations with Type 1 Diabetes or Gestational Diabetes: Systematic Review and Meta-Analysis

Efficacy and Safety of Rapid-Acting Insulin Analogs in Special Populations with Type 1 Diabetes... Diabetes Ther (2018) 9:891–917 https://doi.org/10.1007/s13300-018-0411-7 REVIEW Efficacy and Safety of Rapid-Acting Insulin Analogs in Special Populations with Type 1 Diabetes or Gestational Diabetes: Systematic Review and Meta-Analysis . . . Kirsten Nørgaard Nithya Sukumar Snorri B. Rafnsson Ponnusamy Saravanan Received: February 6, 2018 / Published online: April 5, 2018 The Author(s) 2018 suitable studies for review and inclusion in a ABSTRACT meta-analysis. Eligible studies were randomized controlled trials that reported data on relevant Introduction: To assess the efficacy and safety clinical outcomes. A different reviewer abstrac- of three available rapid-acting insulin analogs ted data for each of the three subpopulations, (insulins lispro, aspart and glulisine, respec- and one reviewer abstracted data for all three. tively) in pregnant women, children/adoles- Any differences were resolved by consensus or cents and people using continuous by consulting a fourth reviewer. subcutaneous insulin infusion (CSII) with type Results: In people on CSII, rapid-acting insulin 1 diabetes. analogs lowered postprandial plasma glucose Methods: PubMed, EMBASE and Cochrane post-breakfast to a greater extent than did reg- Reviews were searched electronically, and their ular human insulin (RHI) (mean difference: bibliographies examined to identify - 1.63 mmol/L [95% confidence interval - 1.71; - 1.54]), with a comparable risk of Enhanced content To view enhanced content for this hypoglycemia and a trend for lower glycated article go to https://doi.org/10.6084/m9.figshare. hemoglobin. In the pediatric population, gly- cemic control was similar with rapid-acting Electronic supplementary material The online insulin analogs and RHI, with no safety con- version of this article (https://doi.org/10.1007/s13300- cerns. Meta-analysis indicated severe hypo- 018-0411-7) contains supplementary material, which is glycemic events were comparable for rapid- available to authorized users. K. Nørgaard S. B. Rafnsson Department of Endocrinology, Copenhagen Centre for Primary Health and Social Care, London University Hospital Hvidovre, Kettega ˚ rd Alle ´ 30, Metropolitan University, 166–220 Holloway Road, 2650 Hvidovre, Denmark London N7 8DB, UK K. Nørgaard S. B. Rafnsson Steno Diabetes Center Copenhagen, Niels Department of Epidemiology and Public Health, Steensensvej 2, 2820 Gentofte, Denmark University College London, 1–19 Torrington Place, London WC1E 7HB, UK N. Sukumar  P. Saravanan (&) Diabetes, Endocrinology and Metabolism, Division P. Saravanan of Health Sciences, Warwick Medical School, Department of Diabetes, Endocrinology and University of Warwick, Coventry CV4 7AL, UK Metabolism, George Eliot Hospital, e-mail: p.saravanan@warwick.ac.uk Nuneaton CV10 7DJ, UK 892 Diabetes Ther (2018) 9:891–917 acting insulin analogs versus RHI (risk differ- in the chemical composition of their formula- ence: 0.00 [95% confidence interval - 0.01; tions [10], but their pharmacokinetic (PK) and 0.01]). In the pregnancy group, insulin lispro pharmacodynamic (PD) profiles are similar and insulin aspart were safe and effective for (Electronic Supplementary Material [ESM] both mother and fetus, with glycemic control Table S1). A large body of clinical studies indi- being at least as good as with RHI. There were cates that these three RAIAs have similar effi- no data on insulin glulisine during pregnancy. cacy and safety [11] and that they are preferred Conclusion: Rapid-acting insulin analogs over regular human insulin (RHI) for use in appear generally safe and effective in these adults with T1D due to their lower risk of special populations; however, additional trials hypoglycemia [12]. would be helpful. However, certain subgroups of patients with Funding: Novo Nordisk A/S. T1D (e.g. children and adolescents, pregnant women and people using CSII) are typically excluded from trials conducted for regulatory Keywords: CSII; Pediatrics; Pregnancy; Rapid- approval to ensure a more homogeneous group acting insulin analogs; Type 1 diabetes of patients. A consequence of excluding these patients INTRODUCTION from regulatory trials is that there is some uncertainty about the clinical profile of RAIAs Many people with type 1 diabetes (T1D) receive in patients who may have unique metabolic, developmental, cognitive or behavioral issues insulin therapy [1]. Rapid-acting insulins are typically used to control postprandial plasma that materially affect the suitability of any medication. With RAIAs in widespread clinical glucose (PPG) excursions, whereas long-acting basal insulins are used to control fasting glu- use, it is timely to examine the available evi- dence for their performance in special popula- cose. Rapid-acting insulins are used as part of a basal–bolus injection regimen as well as for tions. Thus, we performed a systematic review and meta-analysis of published data (PROSPERO continuous subcutaneous insulin infusion registration #CRD42016043006). (CSII). Three rapid-acting insulin analogs (RAIAs) are currently available in the USA and Europe: insulin lispro (Humalog ; Eli Lilly, METHODS Indianapolis, IN, USA), insulin aspart (Novolog in the USA and NovoRapid in the Sources of Data and Search Criteria EU; Novo Nordisk, Bagsværd, Denmark) and insulin glulisine (Apidra ; Sanofi Aventis, The search terms ‘‘insulin lispro’’ (MeSH Terms) Bridgewater, NJ, USA). All three of these RAIAs OR ‘‘insulin’’ (All Fields) AND ‘‘lispro’’ (All are also approved for the pediatric T1D patient Fields) OR ‘‘insulin lispro’’ (All Fields) OR ‘‘lis- population, although the ages on the product pro’’ (All Fields) OR aspart (All Fields) OR gluli- inserts for which there are data vary by product sine (All Fields) were used to search the PubMed, as well as by country (e.g. USA: insulin lispro, EMBASE and the Cochrane Reviews databases children C 3 years of age; insulin aspart, electronically on 1 June 2016 to identify records C 2 years; insulin glulisine, C 4 years [2–4]; EU: for further examination. The titles and abstracts insulin lispro, age not specified; insulin aspart, (and, when necessary, full papers) were then C 1 years; insulin glulisine, C 6 years) [5–7]. In screened to identify papers potentially report- addition, despite the concerns in some coun- ing relevant in vivo data on safety or efficacy in tries, the use of RAIAs is very high in T1D randomized controlled trials (RCTs) involving patients in general and is almost 100% among one of three special populations (children and those patients using CSII [8]. adolescents with T1D; pregnant women with The three insulin analogs lispro, aspart and pre-gestational T1D or gestational diabetes glulisine differ in how their molecular structure [GDM]; people with T1D using CSII). We did has been modified from human insulin [9] and Diabetes Ther (2018) 9:891–917 893 not set a date range and therefore included any by consensus or, if necessary, by consulting a study published up to the date of the search. fourth reviewer (PS) not involved in the original Reference lists of retrieved publications and data abstraction process for that population. All targeted review articles of RAIAs were also reviewers are qualified at a post-doctorate level. searched to identify additional records that might be provisionally relevant. Data Synthesis and Statistical Analysis Selection of Studies and Eligibility Criteria For each of the three populations, whenever possible, we extracted data on glycemic control Records identified as provisionally relevant were (HbA1c at baseline and end of trial; percentage then further examined for eligibility to verify of patients achieving HbA1c target[s] as speci- that they were indeed RCTs, either blinded or fied in individual trials; 7- or 8-point self- open-label and of parallel or crossover design, in measured blood glucose [SMBG]); and safety one of the target special populations and that endpoints (e.g. severe, nocturnal, overall hypo- they reported data on one of the identified glycemia; hyperglycemia/ketosis; fetal out- clinical outcomes of interest. Eligibility for the comes). For completeness, we also extracted three study populations was as follows: for the other secondary endpoints if available. Each of pregnancy population, women with either pre- the retrieved studies was assessed for study gestational T1D or GDM; for the pediatric pop- quality, including sample size, reporting of ulation, children or adolescents aged\ 18 years methods, reporting of results and risk of bias. with T1D; for the CSII population, people with Finally, using this information, we graded each T1D of any age using an insulin pump. study according to the following scale (-, ?, The criterion for the treatment in eligible ??), with - indicating poor quality, ? indi- studies was the administration of one of the cating average quality and ?? indicating good three RAIAs compared with either RHI or each quality. other; trials in which the effect of the RAIA All authors reviewed the available outcomes could not be isolated (as in basal–bolus trials in data for each eligible study and decided which which different basal insulins were used in each studies had efficacy and/or safety data that arm) were excluded. We did not pre-specify a would be suitable for data combination and minimum duration for the studies, although meta-analysis within each of the three special very short-term (e.g. 1- to 2-day meal tests populations. Suitability was determined by designed to study PK/PD) were excluded. Out- authors’ assessment of comparability of the come measures of interest included glycated outcomes assessed. hemoglobin (HbA1c), fasting blood glucose Meta-analysis was conducted using Review (FBG) or plasma glucose, PPG after any or all of Manager 5 (RevMan 5.3 [http://community. the three main meals, hypoglycemia, hyper- cochrane.org/tools/review-production-tools/rev glycemia, diabetic ketoacidosis and/or pump/- man-5]) software for Cochrane Reviews. Two catheter occlusion and, for pregnancy, fetal different outcome measures were used for the outcomes. safety variables in the meta-analysis (namely risk difference for severe and any hypoglycemic episodes per month) and three outcome mea- Data Extraction sures for the efficacy variables (namely mean difference in fasting and postprandial blood A standardized data abstraction form was glucose [BG] and in HbA1c). All models were adapted for use in this systematic review. A run as random-effects models. Statistical different reviewer abstracted data for each of the heterogeneity was calculated by using the I three subpopulations (KN, NS, SR) using the statistic, and publication bias was assessed by inclusion and exclusion criteria, and one using a funnel plot and the Egger’s test. reviewer abstracted data for all three of the subpopulations. Any differences were resolved 894 Diabetes Ther (2018) 9:891–917 Compliance with Ethics Statement mixture formulations; outcomes of interest were not reported; PK/PD studies in which basal insu- lins were the focus of the study; study primarily This article is based on previously conducted addressing stability in pumps; health–economics studies and does not involve any new studies of studies without efficacy or safety outcomes; paper human or animal subjects performed by any of written in a language other than English; dupli- the authors. cate record; or study not applicable to this review for other reasons. A flow diagram showing the RESULTS number of retrieved, excluded and included records is shown in Fig. 1. Prior to submission of Our search of the PubMed, EMBASE and the this manuscript (August 2017), PubMed and the Cochrane Reviews databases resulted in the Cochrane Reviews were searched as previously identification of seven eligible studies in preg- described and no additional studies that would nancy, nine in pediatric patients and 13 in people have been provisionally eligible were identified. using CSII. Records were assessed to be ineligible and thereby excluded from the systematic review Pregnancy and meta-analysis for numerous reasons, includ- ing the study not being a RCT (observational Characteristics of the seven eligible studies in studies, letters, case reports or review articles were pregnancy (two in pre-gestational T1D [13–15] excluded); ineligible population; in vitro or ani- and four in GDM [16–19], with two reporting mal studies; study dealt with one of the RAIAs in different outcomes from the same population Fig. 1 Flow diagram showing the number of retrieved, excluded and included records. The dagger () indicates that the continuous subcutaneous insulin infusion (CSII) group included two pediatric CSII studies. The asterisk (*) indicates that in the CSII meta-analyses, the number of studies included varied depending on the outcome assessed. MDI Multiple daily injections, PD pharmacodynamics, PK pharmacokinetics, RCT randomized controlled trial Diabetes Ther (2018) 9:891–917 895 Table 1 Study outcomes: pregnancy First author, Analog HbA1c outcomes BG levels Hypoglycemia Secondary outcomes year of study (n) vs. Definition Results Definition Results (mmol/L) Definition Results Measures Results comparator (n) Hod, 2008 Aspart (157) Tested in early (1) - 0.04% 8-point SMBG performed (1) 6.82 vs. 6.82 Self-reported diary (1) 0.48 (0.20, (1) Mean (± SD) (1) 87.6 ± 12.0 [13] and vs. RHI pregnancy and (- 0.18; for 1 week in each 1.14) overall treatment vs. (2) 6.96 vs. 7.10 (1) Relative risk Mathiesen, (165) end of each 0.11) trimester (p = NS) satisfaction score 83.4 ± 15.3 (95% CI) of major (3) 6.23 vs. 6.48 2007 [14] trimester (p = NS) (p = 0.03) (1) Mean 24-h BG at first nocturnal (2) 0.72 (0.36; (2) Corrected BW (4) NR, p = 0.044 hypoglycemia 1.46) (SEM) (1) Mean difference (2) - 0.08 trimester (2) 3438 (71.5) (95% CI) between (- 0.23; (p = NS) vs. 3555 (72.9) (5) NR, p = 0.153 (2) Mean 24-h BG at (2) Relative risk (3) Preterm birth aspart and RHI at 0.06) (p = 0.09) second trimester (95% CI) of major (6) NR, p = 0.0007 (4) Neonatal hypo end of second (p = NS) hypoglycemia in (3) 20.3% vs. (3) Mean 24-h BG at third trimester (%) aspart vs. RHI 30.6% trimester (2) Mean difference (p = 0.05) (4) Mean difference in at end of third (4) 33.6 vs. 39.7% post-breakfast glucose in trimester aspart vs. RHI in first trimester (5) Mean difference in post-breakfast BG in second trimester (6) Mean difference in post-breakfast BG in third trimester Persson, 2002 Lispro (16) Tested monthly (1) 6.5 Combined values in second (1) 6.50 ± 3.18 vs. Self-reported diary (1) 0 vs. 2 (1) Progression of (1) 18.8 vs. 35.2 [15] vs. RHI (4.8–8.6) and third trimester of 8.56 ± 3.55 retinopathy, % in (1) Median (range) (1) Number of (2) 5.5 vs. 3.9 (2) No difference (17) vs. 6.6 6-point SMBG (p \ 0.01) each group HbA1c at baseline patients (p \ 0.05) in perinatal (4.5–8.6) (6–8 weeks (1) Mean (± SD) post- (2) 0.40 ± 3.20 vs. experiencing severe (2) Neonatal outcomes outcome or gestation) (%) (2) 5.4 breakfast BG (mmol/L) 1.81 ± 3.42 hypo (i.e., anthropometry, neonatal (4.3–5.9) vs. (p \ 0.01) complications, complications (2) Median HbA1c (2) Mean post-breakfast (2) Rate of 5.3 malformations, at 24 weeks glucose increase (3) 0.70 ± 3.16 vs. biochemical hypo (4.7–6.7), symptomatic hypo) gestation 0.60 ± 3.34 (BG \ 3.0) (%) (3) Mean post-lunch p =NS (p = NS) (3) Median HbA1c glucose increase (3) 5.2 before delivery (4) - 0.24 ± 3.10 (4) Mean post-dinner (4.6–5.9) vs. vs. 0.28 ± 2.94 5.0 glucose increase (p \ 0.04) (4.6–6.7), p =NS 896 Diabetes Ther (2018) 9:891–917 Table 1 continued First author, Analog HbA1c outcomes BG levels Hypoglycemia Secondary outcomes year of study (n) vs. Definition Results Definition Results (mmol/L) Definition Results Measures Results comparator (n) Mecacci, 2003 Lispro (25) Tested at diagnosis of (1) - 0.3 ± 9-point SMBG performed (1) post-BF: NR NR (1) Neonates with (1) 12% vs. 37.5% [17] vs. RHI GDM and 0.3 vs. weekly from diagnosis to 5.93 ± 0.62 vs. cranial-thoracic (p \ 0.05) (24) delivery - 0.3 ± 38 weeks gestation 6.84 ± 0.7 circumference (CC/ (2) No difference 0.1 (p \ 0.002) CT) ratio 10-25th (1) Mean difference (1) Mean (± SD) 1-h post- between groups (p = NS) centile, % (± SD) from pre- breakfast BG (2) post-lunch: intervention to 5.91 ± 0.66 vs. (b) Neonatal (2) Mean BG 1-h post- 6.63 ± 0.88 delivery (%) anthropometry (BW, lunch BG ponderal index) and (p \ 0.01) (3) Mean BG 1-h post- complications evening meal BG (3) post-dinner: 6.21 ± 0.49 vs. (4) Mean total BG 1-h 6.71 ± 0.62 postprandial BG (p \ 0.05) (5) Mean total preprandial BG (4) total: 6.02 ± 0.59 vs. (6) Mean total 2-h 6.72 ± 0.73 postprandial BG (p \ 0.01) (5) 4.08 ± 0.45 vs. 4.13 ± 0.73 (p = NS) (6) 5.20 ± 0.62 vs. 5.44 ± 0.69 (p = NS) Di Cianni, Aspart (31) NR NR 5-point SMBG (1) 6.75 ± 1.12 vs. Number of patients 0 vs. 0 vs. 0 (1) Neonatal (1) Higher BW 2007 [16] vs. lispro 6.6 ± 1.05 vs. experiencing any anthropometry: in RHI than (1) Mean (± SD) 1-h post- (33) vs. 7.5 ± 1.3 hypo BW, CC/CT ratio aspart/lispro breakfast BG RHI (32) groups (p \ 0.05) (2) Macrosomia (2) Macrosomia: 9.6%, 12.1%, 15.6% (p = NS) CC/CT ratio lower in RHI than other two groups (p = 0.03) Diabetes Ther (2018) 9:891–917 897 Table 1 continued First author, Analog HbA1c outcomes BG levels Hypoglycemia Secondary outcomes year of study (n) vs. Definition Results Definition Results (mmol/L) Definition Results Measures Results comparator (n) Pettitt, 2007 Aspart (14) Tested at diagnosis of (1) 5.1 ± 0.4 Measured on standardized (1) 4.2 ± 0.57 vs. Self-reported diary (1) 71 (53) vs. (1) Insulin-specific (1) Low insulin- [18] vs. RHI GDM and vs. mixed meal test after 4.8 ± 0.86 69 (23) antibodies specific (1) % of subjects (13) delivery 5.3 ± 0.3 6 weeks antibodies in (2) - 1.09 ± (number of events) (2) 79 (52) vs. (2) Cross-reactive both groups (1) Diagnosis (2) 5.4 (1) Time-adjusted mean 0.55 vs. experiencing 39 (9) antibody binding (± SD) glucose - 0.54 ± 0.74 symptomatic hypo (± SD) at 36/40 (2) 1.4 ± 3.0 vs. (2) Delivery (both (p = 0.003) and 6/52 post- 1.5 ± 2.3 and groups combined) (2) Time-adjusted mean (2) % of subjects partum (%) 2.3 ± 5.4 vs. change in glucose from (3) 4.7 ± 0.19 vs. (number of events) 6.5 ? 13.7 baseline (no insulin) 5.1 ± 0.23 experiencing minor (3) Mean neonatal (p = 0.03) hypo birth weight (± SD) (3) 3.1 (0.5) vs. (3) Mean glucose at 30 min (kg) 3.0 (0.5) time point of meal test (4) 5.4 ± 0.21 vs. (p = NS) 6.2 ± 0.33 (4) Mean birth length (4) Mean glucose at 60 min (p \ 0.01) (± SD) (cm) (4) 49 (2.3) vs. 48 time point of meal test (2.4) (p = NS) Jovanovic, Lispro (19) Tested at baseline Mean 6-point SMBG (1) 5.5 ± 0.30 vs. Self-reported diary (1) 0.65 ± 0.13 (1) Change in insulin- (1) No difference 1999 [19] vs. RHI and 6 weeks after (± SEM) 7.3 ± 0.4 and meter check vs. specific and cross- in maternal (1) Mean (± SEM) post- (23) randomization HbA1c at: (p = NS) 0.93 ± 1.04 reactive response antibody breakfast hyperglycemia (1) Mean (± SEM) (p \ 0.05) from baseline to response (1) Mean (± SEM) (1) 5.47 ± rate (individual patient % (2) 4.5 ± 1.1 vs. pre-breakfast hypo delivery at baseline (%) 0.09 vs. 5.24 of all 6.8 ± 0.86 rate (individual (2) 0.78 ± 0.37 (2) No difference vs. 0.09 readings C 6.67 mmol/ (p = NS) patient % of all vs. (2) Neonatal in neonatal (2) Mean at 6 weeks (p = NS) L), (%) readings \ 3.1 1.98 ± 0.81 anthropometry anthropometry (3) 2.0 ± 0.51 vs. (3) Mean change mmol/L) (%) and outcomes (2) 5.12 ± (2) Mean post-lunch 2.6 ± 0.54 (3) 1.26 ± 0.43 (3) Neonatal glucose from baseline to 6 0.11 vs. hyperglycemia rate (p = NS) (2) Mean pre-lunch vs. (3) No neonatal weeks (% 5.16 ± 0.12 hypo rate 1.43 ± 0.86 hypo- or reduction) (3) Mean post-evening (4) 4.0 ± 0.49 vs. hyperglycemia (3) - 0.35 meal hyperglycemia rate 5.5 ± 0.47 (3) Mean post- (4) 0.88 ± 0.25 (5.7%) vs. (p \ 0.05) evening meal hypo vs. (4) Mean total postprandial - 0.07 rate 2.20 ± 0.86 hyperglycemia rate (2.8%) (4) Mean total (p = 0.002) preprandial hypo rate Aspart Insulin aspart, BF breakfast, BG blood glucose, BW body weight, CC/CT ratio of the cardiac circumference to the thoracic circumference, CI confidence interval, hypo hypoglycemia, GDM gestational diabetes, glulisine insulin glulisine, HbA1c glycated hemoglobin, lispro insulin lispro, NR not reported, NS not significant, RHI regular human insulin, SD standard deviation, SEM standard error of the mean, SMBG self-measured blood glucose 898 Diabetes Ther (2018) 9:891–917 and therefore described in a single row [13, 14]), insulin aspart (rate ratio 0.72 [95% CI 0.36; are summarized in ESM Table S2, and the results 1.46]). of those studies are shown in Table 1. Most (4/6 In a publication reporting additional data on trials) involved \ 50 participants. Study quality pregnancy outcomes by Mathiesen et al. [14], assessment is shown in ESM Table S3. Three Hod et al. [13] indicated that preterm delivery studies were graded as -, two were graded occurred in 20.3 and 30.6% of pregnancies in as ? and one was graded as ??. women receiving insulin aspart and RHI, respectively (p = 0.053). Other secondary pub- lications from these trials indicated that there Women with Pre-Existing T1D were 137 and 131 live births, 14 and 21 fetal Persson et al. compared treatment with insulin losses, and six and nine congenital malforma- lispro with RHI treatment in a group of 33 tions in these groups of women on insulin pregnant women with T1D and found that the aspart and RHI, respectively. Furthermore, BG level was significantly lower after breakfast maternal and cord blood antibody levels for (but not after other meals) with insulin lispro both RHI and insulin aspart remained low for (0.40 ± 3.20 vs. 1.81 ± 3.42 mmol/L; p \ 0.01) both treatments and were similar at 36 weeks [15] (Table 1). The 95% confidence interval (CI) gestation for the 97 women who participated in was not significantly different between treat- the substudy [20]. In a secondary analysis of ments, either at 24 weeks gestation or before data from the same trial, Lloyd et al. reported delivery. The incidence of severe hypoglycemia that these benefits were attained without was low (zero and two episodes for insulin lispro increasing the cost of treatment compared to and RHI, respectively) and there were no dif- RHI [21]. ferences in perinatal outcomes or neonatal complications. The largest trial was an international, paral- Women with GDM lel-group trial enrolling 322 women with T1D The characteristics of four RCTs using RAIAs in who at enrollment were pregnant GDM are summarized in ESM Table S2 [16–19]. for B 10 weeks or planning to become preg- The results indicate that insulin lispro was at nant. Women were randomized to either insu- least as effective as RHI and sometimes lin aspart or RHI, both in combination with demonstrated improved glycemic control (ESM neutral protamine Hagedorn insulin as the basal Table S2) [16, 17, 19]. Insulin aspart was asso- insulin [14]. HbA1c levels were comparable in ciated with significantly lower post-meal BG the two groups at the end of the second and compared with RHI [16, 18]. In the single trial third trimesters (treatment difference, insulin involving a head-to-head comparison of insulin aspart–RHI: - 0.04% [95% CI - 0.18; 0.11], aspart and insulin lispro, mean 1-h post-break- - 0.4 mmol/mol [95% CI - 2.0; 1.2]; and fast BG was similar for the two products - 0.08% [95% CI - 0.23; 0.06], - 0.9 mmol/- (6.75 ± 1.12 vs. 6.6 ± 1.05 mmol/L, respec- mol [95% CI - 2.5; 0.7], respectively). Mean tively) [16]. plasma glucose levels at 90 min post-breakfast In Mecacci et al. [17], hypoglycemia was not were significantly lower in those women reported and, in another trial, there were no receiving insulin aspart arm than in those hypoglycemic events reported for insulin receiving RHI (p = 0.044 and p = 0.001 for end aspart, insulin lispro or RHI [16]. In a study of of first and third trimesters, respectively). The women (n = 27) using insulin aspart or RHI, the mean PPG increment across all meals was lower reported percentage of participants experienc- for the insulin aspart arm than for the RHI arm ing symptomatic hypoglycemic events was at the end of the first and third trimesters (es- similar for both treatments (71 vs. 69%), but timated treatment difference: –0.75 [95% CI more participants using insulin aspart reported - 1.25; - 0.25], p = 0.003 and - 0.40 [95% CI minor hypoglycemia (79 vs. 39%) [18]. In the - 0.80; - 0.01], p = 0.044, respectively). The latter case, this was largely due to two partici- risk of major hypoglycemic events was numer- pants being prone to hypoglycemia. Neonatal ically lower, but not significantly different, for Diabetes Ther (2018) 9:891–917 899 outcomes (weight, length, physical exam) were insulin lispro, respectively; p = 0.039). The fre- good for both insulin aspart and RHI. quencies of hypoglycemic events and other adverse events were similar. Meta-Analysis The pregnancy studies were deemed to be too Meta-Analysis heterogeneous or to lack relevant information Data on glycemic control were deemed to be too for meta-analysis of either efficacy or safety heterogeneous or to lack relevant information outcomes for any of the three RAIAs. for meta-analysis, but the data on severe hypo- glycemic outcomes were able to be combined from five studies (Fig. 2a). Overall, the number Children and Adolescents of severe hypoglycemic events was low (total of 14), and a random-effects model using generic There were nine eligible studies involving inverse variance showed no difference in risk of pediatric patients with T1D [22–30]. The char- severe hypoglycemia with insulin analog treat- acteristics of these studies are presented in ESM ment, risk difference 0 (95% CI - 0.01; 0.01). Table S4 and the results are shown in Table 2. Funnel plots suggested that there was no pub- Three trials compared insulin aspart with RHI lication bias (data not shown). [22, 23, 28], five compared insulin lispro with RHI [24–27, 30] and one compared insulin lis- Patients Treated with CSII pro with insulin glulisine [29]. Most (5/9; 55.5%) trials involved \ 50 participants. Study quality assessment is shown in ESM Table S5. The largest number of eligible studies (n = 13) Four studies were graded as -, two were graded was identified for people using CSII [31–43], of as ? and three were graded as ??. Trials using which two were carried out in pediatric popu- RAIAs in CSII in children are discussed in the lations [42, 43] (ESM Table S6). The results are section ‘‘Patients Treated with CSII’’. presented in Table 3, and the study quality Overall, glycemic control (either HbA1c or assessment is shown in ESM Table S7. Two PPG) with insulin lispro or insulin aspart was studies were graded as -, three were graded equivalent to or better than that with RHI. This as ? and seven were graded as ??. One of the was also true with respect to incidence of studies consisted of two substudies, one of hypoglycemic episodes or other adverse events. which was graded as ?, and the second was There were no head-to-head trials comparing all graded as ?? due to the double-blind three RAIAs in pediatric participants. However, component. a large (n = 572), open-label, parallel-group, Eight studies compared an RAIA with RHI non-inferiority trial compared insulin glulisine and all involved insulin lispro [32, 34–38, 41], with insulin lispro [29]. Insulin glulisine was with one being a pediatric trial [42]. All were demonstrated to be non-inferior to insulin lis- crossover trials of 1–4 months’ duration. All pro (treatment difference in HbA1c: - 0.06% indicated that insulin lispro was associated with [95% CI - 0.24; 0.12]; - 0.7 [95% CI - 2.6; improved glycemic control (HbA1c) and an 1.33] mmol/mol). More children achieved incidence of hypoglycemic events that was American Diabetes Association (ADA) age- similar to or lower than RHI. specific HbA1c targets (at the time of the Three studies involved head-to-head com- study: \ 6 years, HbA1c [ 7.5 to%\ 8.5% [[ 58 parisons of insulin lispro versus insulin aspart to \ 69 mmol/mol]; 6–12 years, HbA1c \ 8.0% [33, 39, 43], with an additional trial also com- [\ 64 mmol/mol]; 13–17 years, HbA1c \ 7.5% paring RHI [31]. The largest of the three trials [\ 58 mmol/mol]; currently, the ADA recom- was a 16-week, open-label RCT in 298 subjects mends HbA1c \ 7.5% [\ 58 mmol/mol] across with T1D aged 4–18 years [43] (ESM Table S6). all pediatric age groups [1]) with insulin gluli- At 16 weeks, the HbA1c in subjects receiving sine than with insulin lispro (overall popula- insulin aspart was deemed to be non-inferior to tion: 38.4 vs. 32.0% for insulin glulisine and the HbA1c in those receiving insulin lispro, and 900 Diabetes Ther (2018) 9:891–917 Table 2 Study outcomes: pediatrics First Analog HbA1c outcomes BG levels Hypoglycemia Secondary outcomes Author, (n) vs. Definition Results Definition Results (mmol/L) Definition Results Measures Results year comparator (n) Philotheou, Glulisine Tested at (1) ? 0.10 ± 0.08 3-point SMBG at (1) 8.77 ± 0.21 vs. Reported from month 4 to (1) 0.06 ± 0.24 vs. Mean (± SD) 2.53 ± 0.68 vs. 2011 (275) vs. baseline and vs. ? endpoint 9.46 ± 0.21 treatment end 0.07 ± 0.27 increase total 4.91 ± 0.65 [29] lispro endpoint 0.16 ± 0.07 (p = 0.014) daily dose of (p = 0.007) (1) Adjusted (1) Severe hypo (± SD): (2) 0.21 ± 0.50 vs. (295) insulin from (1) Adjusted (2) - 0.06 mean (2) 9.76 ± 0.24 vs. BG \ 2.0 mmol/L and 0.20 ± 0.80 baseline, mean change (- 0.24; 0.12) (± SEM) of 9.80 ± 0.23 third-party assistance (3) 3.10 ± 4.33 vs. units/day (± SD) from pre-breakfast (p = 0.894) required or prompt (3) 38.4 vs. 32.0 2.91 ± 4.35 baseline to BG (p value for recovery following glucose (p = 0.0039) (3) 9.20 ± 0.22 vs. endpoint difference treatment, number of 9.04 ± 0.21 between episodes per patient- (2) Difference (p = 0.564) groups) month between treatments in (2) Adjusted (2) Nocturnal hypo adjusted means mean (3) Any hypo (95% CI) for (± SEM) of difference pre-main meal from baseline BG to endpoint (3) Adjusted (3) % achieving mean ADA age- (± SEM) of specific HbA1c 2-h post main targets at meal endpoint Pan´kowska, Aspart (20) Tested at (1) 7.4 ± 0.9 vs. 24-h glycemic (1) 219.8 (12.8) vs. Self-reported by parent (1) 0.1 vs. 0.0 Treatment 5.6 vs. 4.4 2010 vs. RHI baseline, 7.6 ± 1.1 control 211.8 (10.9) during CGM period satisfaction: SD (p = 0.04) (1/20) vs. (0/21) [28] (21) midpoint and measured (p = 0.55) values for mean (2) 7.6 ± 0.9 vs. (1) Severe hypo: endpoint using CGMS (2) 18 vs. 20 change in 7.6 ± 1.0 (2) 4.0 vs. 4.0 BG \ 2.8 mmol/L for 72 h at treatment (1) Mean (p = NS) accompanied by CNS 18/20 vs. 19/21 endpoint satisfaction HbA1c symptoms requiring (3) 1.1 vs. 1.4 score at end of (± SD) at (1) Mean external help, number of study 13 weeks after (± SD) of area episodes per patient-year treatment under glucose of exposure curve, mmol/ (2) Mean (2) Minor hypoglycemia: h/L HbA1c at BG \ 2.8 mmol/L that 26 weeks after (2) Difference were asymptomatic or treatment between self-treatable maximum and (3) Symptoms-only minimum hypoglycemia glucose levels over 24 h Diabetes Ther (2018) 9:891–917 901 Table 2 continued First Analog HbA1c outcomes BG levels Hypoglycemia Secondary outcomes Author, (n) vs. Definition Results Definition Results (mmol/L) Definition Results Measures Results year comparator (n) Cherubini, Aspart (NR) Tested at (1) 7.5 ± 0.8 to 7-point SMBG (1) 7.38 ± 1.0 vs. Self-reported (1) 0.045 vs. 0.035 NR NR 2006 vs. RHI baseline and 7.0 ± 0.4 vs. during study 7.94 ± 1.06 (p = 0.209) (1) Severe hypo: [22] (NR) 6-weekly 7.5 ± 1.4 to period (p = 0.175) BG \ 2.8 mmol/L, (2) 0.214 vs. 0.18 during study 7.4 ± 0.5 (1) Mean 2-h (2) number of episodes per (p = 0.117) (p = 0.018) (1) Change in postprandial Analog \ RHI, patient per day HbA1c BG p = 0.012 (2) Any hypo: (± SD) over (2) Mean fasting (3) 8.89 ± 2.89 vs. BG \ 3.9 mmol/L the study BG 9.17 ± 2.72 period (p = 0.113) (3) Mean afternoon BG (4) Analog = RHI, at endpoint p [ 0.237 (4) Decrease in mean daily BG variability from baseline Fairchild, Lispro (35) Tested at (1) 8.33 ± 0.89 7-point SMBG (1) 10.57 ± 0.26 Self-reported and BG meter (1) 0.032 vs. 0.065 Treatment 28 (80) vs. 7 (20) 2000 vs. RHI baseline and vs. 8.14 ± 0.77 done weekly vs. 9.02 ± 0.46 analysis (p = NS) satisfaction [25] (35) 6-weekly (p = NS) or 2-weekly (p = 0.001) questionnaire (1) Severe hypos: hypo 1/35 vs. 2/35 during study (1) Mean BG (2) 2.35 associated with Number (%) (2) 13.47 vs. 10.77 (1) Mean (± SD) at (-3.98; - 0.72) convulsion or coma preferring this (p = NS) HbA1c 03:00 h type of insulin p = 0.01 (2) Total recorded hypos, (± SD) at (3) 5.69 vs. 3.31, over the other (2) Mean number of episodes per endpoint difference 2.4 ± 5.1 one difference patient per 3 months (p = 0.02) (95% CI) in (3) Total recorded hypos 03:00 h BG (4) analog = comparator from 06:00 to 12:00 h between groups (p = NS) (4) Hypos with BG \ 3.0 mmol/L 902 Diabetes Ther (2018) 9:891–917 Table 2 continued First Analog HbA1c outcomes BG levels Hypoglycemia Secondary outcomes Author, (n) vs. Definition Results Definition Results (mmol/L) Definition Results Measures Results year comparator (n) Ford- Lispro (23) Tested at (1) 8.9 ± 0.3 vs. Overnight (1) 6.5 ± 1.0 vs. Self-reported (1) 2/23 vs. 1/23 Mean (range) total 0.97 (0.68–1.26) Adams, vs. RHI baseline and 8.4 ± 0.3 metabolic 7.1 ± 1.1 daily insulin vs. 0.96 (1) Severe hypo: (2) 1.6 ± 0.3 vs. 2003 (23) 4-monthly study (p = 0.5) dose, units/kg (0.53–1.22) (p = 0.14) convulsions or requiring 1.7 ± 0.3 (p = 0.2) [26] during study (p = 0.2) (1) Mean BG (2) 6.1 ± 0.8 vs. glucagon, number of (2) 8.5 ± 0.2 vs. (3) 9 vs. 6 (p = 0.06) (1) Mean (± SD) at start 6.3 ± 0.9 episodes 8.8 ± 0.3 HbA1c of overnight (p = 0.8) (4) 8 vs. 13 (p = 0.01) (2) Symptomatic (± SD) at (p = 0.47) profile (3) 138 ± 12 vs. hypoglycemia, number of (5) 27 vs. 22 crossover (2) Mean fasting 170 ± 13 episodes per patient per (p = 0.11) (2) Mean BG (p = 0.03) week (± SD) HbA1c at (3) AUC of BG (4) 158 ± 13 vs. Overnight metabolic profile endpoint (± SD) from 145 ± 12 (3) Prevalence of low BG post-EM to (p = 0.3) from post-EM to bedtime, 22:00 h: mmol/min/L BG \ 3.5 mmol/L, % (4) AUC of BG (4) Prevalence of low BG from 22:00 to from 22:00 to 04:00 h 04:00 h (5) Prevalence of low BG from 04:00 to 07:00 h Holcombe, Lispro (457) Tested at (1) 8.41 ± 1.4 vs. 8-point BG (1) 9.7 ± 3.9 vs. Self-reported (1) 5 vs. 5 (1) Mean (± SD) (1) 1.08 ± 0.32 2002 vs. RHI baseline and 8.80 ± 1.5 profiles on 8.8 ± 3.7 total daily vs. 1.05 ± 0.30 (1) Severe hypo: needing 5/457 vs. 5/547 [27] (457) 2-monthly 2 days at (p \ 0.001) insulin dose, (p \ 0.001) (p = NS) third-party assistance or during study baseline and (2) 1.0 ± 1.9 vs. units/kg (2) 10.2 ± 3.5 vs. intravenous glucose or (2) 0.54 ± 0.24 (2) 8.69 ± 1.52 end of each 1.7 ± 2.6 (p \ 0.001) (1) Mean 9.6 ± 3.4 glucagon injection, (2) Mean (± SD) vs. 0.53 ± 0.20 vs. 8.70 ± 1.65 treatment HbA1c (p = 0.005) number of patients (3) 4.02 ± 4.5 vs. daily dose of (p = NS) (p = NS) period (± SD) at 4.37 ± 4.5 short-and rapid- (3) 9.7 ± 4.0 vs. (2) Nocturnal baseline (1) Mean (p = 0.023) acting insulin 10.6 ± 4.3 hypoglycemia: from (± SD) BG (2) Mean (p \ 0.001) midnight to 06:00 h concentrations HbA1c at at 03:00 h (4) 8.6 ± 3.5 vs. (3) Any hypoglycemia: endpoint 9.3 ± 3.7 symptoms present or (2) Mean fasting (p = 0.003) measured BG BG \ 3.0 mmol/L, concentrations (5) 8.1 ± 3.4 vs. number of episodes per 8.5 ± 3.4 (3) Mean BG 2-h patient per month (p = NS) post-breakfast (4) Mean BG 2-h post-dinner (5) Mean BG 2-h post-mid-day meal Diabetes Ther (2018) 9:891–917 903 Table 2 continued First Analog HbA1c outcomes BG levels Hypoglycemia Secondary outcomes Author, (n) vs. Definition Results Definition Results (mmol/L) Definition Results Measures Results year comparator (n) Deeb, 2001 Lispro before Tested at (1) 8.40 ± 1.1 vs. 7-point SMBG (1) 9.6 ± 1.6 vs. Self-reported (1) 2 vs. 3 vs. 6 Mean total daily 0.21 vs. 0.23 vs. [24] meals baseline and 8.54 ± 1.0 vs. profiles on 10.1 ± 1.6 vs. insulin dose, 0.25 (p = NS) (1) Severe hypo: needing (p = NS) (analog 1, 3-monthly 8.43 ± 1.0 2 days at 10.0 ± 1.7 units/kg third-party assistance, 53) vs. during study (p = NS) baseline and 5/108 vs. 6/57 (p = 0.007 analog resulting in a coma, or lispro after end of each (1) Mean 1 vs. analog 2 requiring intravenous (2) 14.7 ± 11.9 vs. meals treatment (± SD) (p = 0.024 for glucose or glucagon, 13.6 ± 9.3 vs. (analog 2, period HbA1c at trend) number of patients 13.8 ± 9.8 (p = NS) 55) vs. endpoint (1) Overall mean RHI (57) (2) 11.7 ± 4.4, (2) Any hypoglycemia: BG (± SD) 13.5 ± 5.5, symptomatic or measured (2) Mean BG 2-h 15.0 ± 5.4 BG \ 3.5 mmol/L, post-breakfast (p \ 0.001 number of episodes per analog 1 vs. 30 days (3) Mean BG comparator; pre-lunch p = 0.023 analog 1 (4) Mean BG 2-h vs. analog 2) post-lunch (3) 8.7 ± 3.9 vs. (5) Mean BG 2-h 8.3 ± 3.1 vs. post-dinner BG 8.3 ± 3.1 vs. 9.5 ± 4.1 (p = NS analog 1 vs. analog 2, p = 0.037 analog 1 vs. RHI) (4) Analog 1 = analog 2 = RHI (p = NS) (5) 8.8 ± 5.0 vs. 9.9 ± 4.7 vs. 10.8 ± 5.4 (p = NS analog 1 vs. analog 2, p = 0.006 analog 1 vs. RHI) 904 Diabetes Ther (2018) 9:891–917 Table 2 continued First Analog HbA1c outcomes BG levels Hypoglycemia Secondary outcomes Author, (n) vs. Definition Results Definition Results (mmol/L) Definition Results Measures Results year comparator (n) Tupola, Lispro (22) Tested at (1) 0.2 ± 0.8 vs. 7-point SMBG (1) 11.5 ± 4.5 vs. Self-reported and BG meter (1) 2 vs. 2 (p = NS) (1) Total daily (1) Analog = RHI 2001 vs. RHI baseline and - 0.4 ± 0.7 profiles on 8.4 ± 3.8 analysis insulin dose (p = NS) (2) 34 vs. 41 (p = 0.6) [30] (22) 3-monthly 1 day per (p = 0.02) (p = 0.1) (1) Severe hypo: loss of (2) Patient (2) 18 (82) vs. 4 during study month over (3) 4.9 vs. 4.4 (p = 0.3) (2) 11.7 ± 6.0 vs. consciousness during a satisfaction: (18) study period (1) Mean 9.6 ± 5.7 hypoglycemic episode number (%) of (± SD) (1) Mean BG (p = 0.4) patients/families (2) Nocturnal change in (± SD) pre- who wanted to (3) 11.5 ± 5.0 vs. hypoglycemia: from 11:00 HbA1c from breakfast continue this 10.6 ± 6.0 to 06:00 h baseline to type of insulin (2) Mean BG (p = 0.8) endpoint (3) Hypoglycemia: pre-dinner (4) Analog = RHI symptoms present or (3) Mean BG at (p = NS) BG \ 3 mmol/L, bedtime number of episodes per patient per month (4) 1-h and 2-h postprandial BG excursions Danne, Aspart (23) Tested at (1) Analog = RHI 7-point SMBG (1) Analog = RHI Self-reported (1) 2 vs. 3 WHO–DTSQs (1) 4.8 ± 1.5 vs. 2007 vs. RHI baseline and (p = NS) during study (p = NS) 3.7 ± 1.8 (1) Major hypoglycemia: (2) 2.75 vs. 2.74 (1) Mean score [23] (25) 3-monthly period those that parents could (± SD) for (p = 0.045) during study (3) 1.06 (0.96;1.17) (1) Daily BG not handle on their own, question on (p = 0.225) (2) 5.0 ± 1.4 vs. (1) HbA1c at variations number of episodes continuing 4.2 ± 1.8 endpoint present form of (2) Any hypoglycemia, insulin (p = 0.051) number of episodes per week (2) Mean score (± SD) for (3) Relative risk of question on hypoglycemia (95% CI): recommending relative risk for analog/ this form of RHI insulin to others ADA American Diabetes Association, AUC area under the curve, CGM continuous glucose monitoring, CNS central nervous system, DTSQ Diabetes Treatment Satisfaction Questionnaire, DKA diabetic ketoacidosis, EM evening meal, WHO World Health Organization Diabetes Ther (2018) 9:891–917 905 906 Diabetes Ther (2018) 9:891–917 bFig. 2 Meta-analyses of key outcomes. a Forest plot there were no significant differences in FBG or showing the difference in risk of severe hypoglycemic rates of hyper- and hypoglycemia. However, the episodes with insulin analog treatment compared to regular daily insulin dose was significantly lower for human insulin (RHI) in a pediatric population. b Forest groups on insulin aspart (0.86 ± 0.237 vs. plot showing the difference in the mean fasting blood 0.94 ± 0.233 U/kg, for insulin aspart vs. insulin glucose level with insulin analog treatment compared to lispro, respectively; p = 0.018) [43]. In two RHI treatment in the CSII sub-review. c Forest plot related, 24-week, randomized, crossover trials in showing the difference in mean postprandial blood glucose adults with T1D, insulin lispro was assessed to (BG) level with insulin analog treatment compared to RHI be non-inferior to insulin aspart based on SMBG treatment in the CSII sub-review. d Forest plot showing profiles averaged over days 1–6 of treatment, the difference in risk of severe hypoglycemic episodes with but not when day 6 values alone were compared insulin analog treatment compared to RHI treatment in [39]. In a randomized, parallel-group trial in 146 the CSII sub-review. e Forest plot showing the mean adults with T1D, mean change from baseline difference in the rate of any hypoglycemic episodes with was not significantly different for participants insulin analog treatment compared to RHI treatment in treated with insulin lispro, insulin aspart or RHI the CSII sub-review. f Forest plot showing the difference for 16 weeks [31]. Rates of hypoglycemia were in glycated hemogloblin (HbA1c) with insulin analog also similar among treatments. treatment (lispro or aspart) compared to RHI in the CSII One RCT compared all three RAIAs in adults sub-review. [40]. This was a crossover trial with three Squares and diamonds represent the difference in HbA1c 13-week periods that was designed to test the after intervention with the two treatments for each study superiority of insulin glulisine for unexplained (horizontal lines are 95% CI) and for all the studies combined, respectively. The I value refers to the statistical hyperglycemia and/or infusion-set occlusion. It heterogeneity for the pooled analysis. A random-effects failed to show superiority of insulin glulisine on model using generic inverse variance showed a mean the primary outcome, but revealed that the difference in HbA1c of - 0.19% (95% CI - 0.46; 0.08); monthly rate of unexplained hyperglycemic - 2.1 (95% CI - 5.0; 0.9) mmol/mol with insulin analog episodes and/or perceived catheter-set occlu- compared to RHI at the end of the treatment period. The sion was significantly higher in insulin gluli- squares and the diamond in a, d, e represent the difference sine-treated patients than in those receiving the in risk for each study (horizontal lines represent 95% CI) two other analogs [40]. Furthermore, insulin and for all studies combined, respectively. The squares and glulisine was associated with a higher frequency the diamond in b, c represent the difference in the glucose of symptomatic hypoglycemia, whereas HbA1c levels between the two treatment arms for each study and 7-point SMBG were similar for all three (horizontal lines are 95% CI) and for all the studies insulin analogs [40]. combined, respectively. The results of these meta-analyses are the mean of post-breakfast BG measurements only. In Meta-analysis a and d ‘Events’ refers to the number of patients Data on mean FBG and mean PPG for patients experiencing any such episode during the treatment period using CSII were sufficiently consistent to permit as a proportion of total number of patients in that some meta-analysis, as were some of the hypo- treatment group. ‘Rate’ refers to mean (± SD) of any glycemic outcomes. Funnel plots suggested that episodes of hypoglycemia per 30 days in all the patients in there was no publication bias, although the the respective treatment group. In f ‘Bode, 2002 (a)’ [31] refers to the observed difference in HbA1c between the number of studies in the meta-analysis with subgroup of insulin lispro vs. RHI, and ‘Bode, 2002 (b)’ FBG outcomes was small (funnel plots not [31] refers to the subgroup on insulin aspart vs. RHI; the shown). Meta-analysis was performed for RAIAs three remaining studies compare lispro vs. RHI. The ‘I ’ versus RHI for FBG (three studies) (Fig. 2b), PPG value refers to the statistical heterogeneity for this pooled (five studies) (Fig. 2c), severe hypoglycemic analysis. CI confidence interval, IV inverse variance, episodes (six studies) (Fig. 2d), any hypo- SD standard deviation glycemic episodes (five studies) (Fig. 2e) and HbA1c (four studies) (Fig. 2f). A random-effects model using generic inverse variance showed a mean difference in FBG of - 0.53 mmol/L Diabetes Ther (2018) 9:891–917 907 Table 3 Study outcomes: continuous subcutaneous insulin infusion First author, Analog HbA1c outcomes BG levels Hypoglycemia Secondary outcomes year (n) vs. Definition Results Definition Results (mmol/L) Definition Results Measures Results comparator (n) Zinmann, Lispro (30) Tested monthly (1) 8.03 ± 0.13 8 point-SMBG (1) No significant Self-reported diary (1) 12.7 ± 1.6 (1) 5-h meal test on (1) Peak free plasma insulin at 1997 [41] vs. RHI profile done difference between subgroup of six 45 min for lispro (1) Mean (2) 7.66 ± 0.13 vs. (1) Any hypo at (2) 8.6 ± 1.4 vs. (30) weekly treatments patients for (287 ± 69 pmol/L) vs. (± SE) 8.00 ± 0.16 baseline: 10.8 ± 1.8 postprandial 150 min for RHI (294 ± 56). baseline (1) Mean (2) Lower with lispro than BG \ 3 mmol/L or (lispro (3) - 0.34 insulin glucose Plasma glucose and FFA HbA1c in all fasting BG with RHI (p \ 0.05) compatible p = 0.035 vs. (p = 0.0041) and FFA markedly reduced for lispro vs. patients symptoms, episodes baseline, RHI (2) Mean post- RHI per 30 days (± SE) p = NS) (2) BW (2) Mean prandial BG (2) No difference in BW from endpoint (2) Any hypo during (3) 8.4 ± 1.3 (3) Infusion-set baseline or between treatments HbA1c treatment occlusion (4) 6.0 ± 0.9 vs. at endpoint (3) Difference in (3) Biochemical hypo 7.6 ± 1.3 (3) No occlusions with either endpoint at baseline: (lispro treatment HbA1c BG \ 3 mmol/L p = 0.03 vs. between only baseline, RHI treatments p = NS) (4) Biochemical hypo during treatment (5) 0/30 vs. 0/30 (5) Severe hypo Melki, 1998 Lispro (38) Tested at end of (1) 7.74 ± 0.20 vs. 7 point-SMBG (1) 7.93 ± 0.15 vs. Self-reported diary (1) 7.03 ± 0.94 (1) Ketoacidosis (1) No episodes in either group [35] vs. RHI first 7.97 ± 0.13 profile done 8.61 ± 0.18 recorded in last vs. (2) BW (2) No difference (38) 3 months daily during (p \ 0.0001) 30 days of first 7.94 ± 0.88 (2) 7.11 ± 0.15 vs. last 30 days treatment period (p = NS) (3) Glucose (3) Significantly lower overall (1) Baseline 7.88 ± 0.16 (2) 7.70 ± 0.17 vs. of each variability mean and postprandial glycemic 7.75 ± 0.21 (p = NS) (1) Hypo event: (2) 0.05 ± 0.05 (2) Endpoint (3) - 0.62 ± 0.13 treatment fluctuation with lispro than BG \ 3 mmol/L, vs. 0.47 ± 0.19 (4) Patient vs. - 0.09 ± period (3) 8.26 ± 0.19 vs. RHI, SD of BG in mmol/L (3) Change from episodes per month (p \ 0.05) preference 0.15 (p = 0.01) 9.90 ± 0.20 (± SE): 3.44 ± 0.10 vs. baseline to (1) Mean BG (± SD) (p \ 0.0001) (3) 3 vs. 7 (3/38 (5) Technical 3.80 ± 0.10 (p \ 0.001) and endpoint ± SE (2) Very low BG vs. 4/38) problems 3.58 ± 0.10 mmol/L vs. (2) Mean measurement: 3.84 ± 0.10 (p \ 0.02) preprandial BG \ 2 mmol/L (4) All seven questions on non- BG (3) Severe hypo: third- validated questionnaire in favor (3) Mean 2-h party assistance of lispro (p \ 0.0001) postprandial required (5) Insulin precipitation in BG catheter, one vs. four episodes; catheter obstruction, nine episodes each 908 Diabetes Ther (2018) 9:891–917 Table 3 continued First author, Analog HbA1c outcomes BG levels Hypoglycemia Secondary outcomes year (n) vs. Definition Results Definition Results (mmol/L) Definition Results Measures Results comparator (n) Schmauss, Lispro (11) Tested Lispro then RHI vs. 3-day BG (1) 6.5 ± 0.4 vs. Self-reported (1) 4.0 ± 0.9 vs. (1) Basal and bolus (1) No difference 1998 [38] vs. RHI 3-monthly RHI then lispro profile done 7.5 ± 0.4 3.2 ± 0.7 insulin (1) Hypo: (2) No change in BMI (11) 3-monthly (p = NS) requirements (1) Baseline (1) 6.3 ± 0.2 vs. (p = NS) BG \ 3.5 mmol/L (3) No difference in treatment 6.7 ± 0.4 (1) Mean and/or symptoms, (2) 0 vs. 0 (0/11 (2) BMI change (2) End of first (2) 6.8 ± 0.3 vs. satisfaction fasting BG episodes per 30 days vs. 0/11) 3 months (p = NS) 8.3 ± 0.3 (3) Treatment (± SD) (2) Mean 2-h satisfaction (3) End of study (2) 5.7 ± 0.3 vs. (p = 0.03) postprandial (2) Severe hypo: 6.5 ± 0.3 BG requiring IV glucose (p = NS) or glucagon (3) 6.2 ± 0.2 vs. 6.3 ± 0.3 (p = NS) Guerci, 1999 Lispro (10) Tested monthly (1) 7.17 ± 0.86 vs. SMBG done (1) 9.35 ± 1.17 vs. Self-reported (1) 10.1 ± 9.7 vs. (1) Glucose (1) No difference between [32] vs. RHI 7.36 ± 0.76 monthly 9.07 ± 0.43 (p = NS) 6.9 ± 4.4 variability, mean treatment groups (1) Baseline (1) Incidence of hypo (10) (p = NS) (p = NS) SD of previous (1) Mean (2) 9.04 ± 0.89 vs. at baseline: (2) Until 3 h no difference in PG (2) Endpoint of months’ BG (2) 7.07 ± 0.51 vs. baseline BG 9.32 ± 1.17 (p = NS) BG \ 3.5 mmol/L, (2) 7.1 ± 4.6 vs. between groups. From 3 to 5 h, each treatment 6.97 ± 0.67 episodes per 30 days 12.6 ± 10.2 After pump PG higher with lispro (p \ 0.01) period (2) Mean BG (3) 9.53 ± 1.98 vs. (p = NS) (± SD) (p = 0.05) interruption for at endpoint 9.92 ± 1.05 (p = NS) (3) From 3 h onwards, consistently 5h: (2) Incidence of hypo higher for lispro vs. RHI (but (3) Mean (4) 9.43 ± 1.39 vs. at endpoint (2) Mean PG p = NS) baseline 10.49 ± 2.05 postprandial (p = 0.05) (3) Plasma (4) From 2 h onwards, consistently BG 3-hydroxybuturate higher for lispro vs. RHI (p \ 0.05) (4) Mean (4) Plasma FFA postprandial BG at endpoint Diabetes Ther (2018) 9:891–917 909 Table 3 continued First author, Analog HbA1c outcomes BG levels Hypoglycemia Secondary outcomes year (n) vs. Definition Results Definition Results (mmol/L) Definition Results Measures Results comparator (n) Renner, 1999 Lispro (113) Tested at (1) 7.24 ± 1.0 8-point daily (1) 7.2 ± 1.7 vs. Self-reported (1) 12.0 ± 13.9 (1) Ketosis (1) Five vs. four patients [37] vs. RHI baseline and SMBG 7.8 ± 2.1 vs. (2) 6.77 ± 0.88 vs. (1) BG \ 3.5 mmol/L (2) Treatment (2) Patients scored significantly (113) endpoint 11.0 ± 11.2 6.90 ± 0.97 (1) Mean (p = NS) and/or symptoms, satisfaction higher on the treatment (p = NS) (1) Baseline in (p = 0.02) fasting BG mean number of (DTSQ) satisfaction when treated with (2) 7.0 ± 1.9 vs. all patients (± SD) episodes (± SD) per (2) Lispro = RHI lispro compared to RHI 8.6 ± 2.6 (p \ 0.001) (3) Occlusion of patient per 4 months (p = NS) (2) Endpoint (2) Mean post- catheter (3) Same amount in the two (3) 6.9 ± 1.9 vs. breakfast BG (2) BG \ 3.3 mmol/L, treatments 7.3 ± 2.2 mean number of (3) Mean pre- (p = NS) episodes (± SD) per lunch BG patient per month (4) 7.6 ± 1.9 vs. (4) Mean post- 8.7 ± 2.4 lunch BG (p \ 0.001) (5) Mean pre- dinner BG (5) 7.3 ± 1.9 vs. 7.5 ± 1.9 (6) Mean post- dinner BG (p = NS) (7) Mean (p \ 0.001) 22:00 h BG (6) 7.2 ± 1.9 vs. (8) Mean 8.3 ± 1.9 02:00 h BG (7) 7.6 ± 1.8 vs. 8.3 ± 2.0 (p \ 0.001) (8) 8.0 ± 2.7 vs. 7.7 ± 2.3 (p = NS) Johansson, Lispro (41) Tested at (1) 7.7 ± 0.8 Last 30 days (1) 8.5 vs. 8.4 (p = NS) Self-reported (1) 9.7 vs. 8.0 [1.7 (1) Treatment (1) No difference between the 2000 [34] vs. RHI baseline and SMBG (- 1.3; 5.3)] satisfaction treatments (2) 7.4 vs. 7.6 (2) 8.1 vs. 9.6 (p \ 0.001) (1) Any hypo: (41) endpoint (p = NS) (DTSQ) (1) Mean of BG \ 3.0 mmol/L (p = 0.047) (3) 8.3 vs. 8.9 (p \ 0.001) (1) Baseline in pre-prandial and/or symptoms, all patients (3) - 0.2 (- 0.3; and bedtime episodes per 30 days 0.0) BG (difference between (2) Endpoint treatments [95% (p = 0.047) (2) Mean of (3) Difference CI]) postprandial between BG treatments (95% CI) (3) Mean of all SMBG 910 Diabetes Ther (2018) 9:891–917 Table 3 continued First author, Analog HbA1c outcomes BG levels Hypoglycemia Secondary outcomes year (n) vs. Definition Results Definition Results (mmol/L) Definition Results Measures Results comparator (n) Raskin, 2001 Lispro (58) Tested at (1) 7.9 ± 1.1 vs. 4-point daily (1) 8.1 ± 2.0 vs. Self-reported (1) 8 vs. 11 (1) BW (1) No difference between [36] vs. RHI baseline and 7.6 ± 0.8 SMBG 8.1 ± 1.6, (p = NS) (p = NS) treatments (1) Any hypo: (58) endpoint (p = NS) done in last (2) 11.16 ± 4.29 vs. BG \ 3.0 mmol/L (2) 3 vs. 3 2 weeks of (1) Baseline in (2) - 0.34 ± 13.20 ± 4.68 and/or symptoms, (p = NS) each all patients 0.59 vs. (p = 0.012) episodes per treatment 3/58 vs. 2/58 - 0.09 ± 12 weeks (2) Change from period (3) 9.64 ± 4.10 vs. 0.63 (p = 0.004) baseline 12.53 ± 4.64 (2) Severe (1) Mean BG (p = 0.001) hypoglycemia: (± SD) requiring IV glucose PG during test meal at end of each treatment period (2) Mean PG 1 h post- meal (3) Mean PG 2 h post- meal Tubiani-Rufi, Lispro (27) Tested at (1) ? 0.15 ± 0.13 9-point SMBG (1) - 0.94 ± 5.05 vs. ? Self-reported (1) 2 vs. 2 (1) BW (1) No difference between 2004 [42] vs. RHI baseline and vs. ? on 2 days 1.78 ± 5.94 (p = 0.01) treatments (1) Severe 2/27 vs. 2/27 (2) Glucose (27) endpoint 0.11 ± 0.63 during each hypoglycemia, variability (SD of (2) 6.4 ± 1.0 vs. (p = NS) treatment (2) 11.0 ± 6.4 vs. (1) Difference in episodes per 30 days SMBG) 6.0 ± 0.95 mmol/L (p = 0.01) 13.8 ± 8.5 HbA1c (1) Change in (2) BG B 3.3 mmol/L (p = NS) (3) Infusion-set (3) No difference between (± SD) from mean post- occlusion treatments baseline to end dinner BG (3) BG B 2.2 mmol/L (3) 0.6 ± 1.1 vs. of first 1.0 ± 1.1 (4) Parents’ (4) 74% want to continue lispro treatment (p = NS) treatment (5) 2.5 ± 2.7 vs. 2.0 ± 3.0 sequence preference episodes (p = NS) (5) Hyperglycemic episodes with ketonuria Diabetes Ther (2018) 9:891–917 911 Table 3 continued First author, Analog HbA1c outcomes BG levels Hypoglycemia Secondary outcomes year (n) vs. Definition Results Definition Results (mmol/L) Definition Results Measures Results comparator (n) Hoogma, Glulisine Tested at (1) 6.8 vs. 7.1 NR NR Self-reported (1) 2 vs. 2 (1) Infusion-set (1) No difference between 2006 [33] (29) vs. baseline and (p = NS) occlusion treatments (1) Severe (2) 20 vs. 15 aspart endpoint (2) 7.0 vs. 7.2 hypoglycemia: (p = NS) (2) Unexplained (2) 6 vs. 12 (p = NS) (30) (1) Baseline BG \ 2.0 mmol/L hyperglycemia: at (3) 0.11 (3) 26 vs. 24 HbA1c and requiring least one episodes (p = NS) (- 0.09; 0.31) assistance or IV of (2) Endpoint (p = NS) glucose/glucagon, BG [ 19.4 mmol/ HbA1c number of patients L, episodes in (3) Between- experiencing at least 12 weeks group one in 12 weeks difference in (2) Nocturnal HbA1c hypoglycemia (95% CI) from baseline (3) Symptomatic to endpoint hypoglycemia Weinzimer, Aspart (198) Tested at (1) 8.0 ± 0.94 vs. 8-point SMBG (1) 9.5 ± 4.3 vs. Self-reported (1) 0.4 vs. 0.3 (1) BW (1) No difference between 2008 [43] vs. lispro baseline and 8.2 ± 0.84 done on 9.9 ± 3.8 (p = NS) (p = NS) treatments (1) Major (2) Hyperglycemia: (100) endpoint (p = NS) 2 days at (2) 9.3 ± 3.7 vs. hypoglycemia: (2) 5.7 vs. 6.2 BG [ 16.7 mmol/ (2) 11 vs. 17 (p = NS) baseline and (1) Baseline (2) - 0.15 ± 10.0 ± 4.6 BG \ 3.1 mmol/L (p = NS) L, % patients with before (3) 1 vs. 2 (p = NS) 0.05 vs. and requiring episodes (2) Change in ending (p = NS) (3) 77.2 vs. 66.0 - 0.05 ± 0.07 assistance or IV HbA1c from treatment (p = NS) (3) DKA, number of (p = NS) (3) aspart = lispro glucose/glucagon, baseline to end patients with (1) Mean (p = NS) episodes per patient of treatment (3) 50.3 vs. 40.4% episodes fasting BG per year (p = NS) (4) aspart = lispro (3) Percentage (± SD) at (p = NS) (2) Nocturnal achieving age- (4) 59.7 vs. 43.8% baseline hypoglycemia: minor specific (p = 0.04) (2) Mean or major occurring HbA1c target fasting BG at between midnight at baseline endpoint and 06:00 h (4) Percentage (3) Mean (3) Minor achieving age- postprandial hypoglycemia: specific BG at BG \ 3.1 mmol/L HbA1c target endpoint with or without at endpoint symptoms (4) Mean BG at endpoint 912 Diabetes Ther (2018) 9:891–917 Table 3 continued First author, Analog HbA1c outcomes BG levels Hypoglycemia Secondary outcomes year (n) vs. Definition Results Definition Results (mmol/L) Definition Results Measures Results comparator (n) Tamborlane, Substudy 1 Tested at Substudy 1 7-point SMBG Substudy 1 Self-reported (1) 5 vs. 16 (1) Infusion-set (1) No difference between 2015 [39] baseline and done on occlusion treatments lispro (116) (1) 7.3 ± 0.7 (1) 9.34 ± 0.14 vs. (1) Severe hypo: (2) lispro = endpoint day 6 of vs. aspart 9.11 ± 0.14 (p [ 0.05) requiring third-party aspart (2) Hyperglycemia: (2) Substudy 1: 8.20 vs. 6.79 (2) 7.34 ± 0.05 vs. reservoir use (116) (1) Mean assistance, total BG[ 13.9 (p = 0.029) 7.28 ± 0.05 Substudy 2 (3) Substudy 1: (± SD) (1) Mean number of episodes mmol/L ([3h Substudy 2 (p = NS) 9.39 vs. 10.84 Substudy 2: 8.05 vs. 6.54 HbA1c at (± SEM) (1) 8.72 ± 0.10 vs. after eating) (2) Nocturnal hypo: (p = 0.003) (p = 0.003) lispro (118) baseline (3) 0.06 ± 0.05 vs. BG 8.54 ± 0.10 (p [ 0.05) or[ 16.7 mmol/L occurring between vs. aspart - 0.00 ± 0.05 Substudy 2: 7.57 (within 3 h after (3) 4 vs. 0 (2) Mean bedtime and wake up (118) (p = NS) vs. 8.71 eating), episodes per (± SEM) (3) Symptomatic hypo: (p = 0.012) 30 days endpoint Substudy 2 BG \ 3.9 mmol/L HbA1c (4) Substudy 1: (3) DKA, number of (1) 7.5 ± 0.7 and symptomatic, 15.26 vs. 16.91 episodes (3) Mean change episodes per 30 days (2) 7.30 ± 0.04 vs. (p = 0.006) (± SEM) in 7.14 ± 0.04 (4) Symptomatic hypo: HbA1c from Substudy 2: 16.74 (p \ 0.001) total number of baseline vs. 18.86 episodes (3) - 0.15 vs. (p \ 0.001) - 0.31 (5) Asymptomatic: (5) BG \ 3.9 mmol/L (p \ 0.001) lispro = aspart and no symptoms Bode, 2002 Aspart (59) Tested at (1) 7.3 ± 0.7 vs. 8-point SMBG (1) Aspart = lispro = Self-reported (1) 0 vs. 0 vs. 1 (1) BW (1) No difference in any treatment [31] vs. lispro baseline and 7.3 ± 0.7 vs. done on RHI group (1) Severe hypo: 0/59 vs. 1/59; (2) Unexplained (28) vs. endpoint 7.5 ± 0.8 2 days at (2) BG \ 2.8 mmol/L 0/28 vs. 1/59 hyperglycemia: (2) 27% vs. 36% vs. 41%, RHI (59) (p = NS) baseline and (1) Mean Aspart \ lispro = RHI and requiring BG [ 19.4 mmol/ endpoint (2) 0.5 ± 0.83 vs. (p = NS) (± SD) (2) 0.00 ± 0.51 vs. assistance or IV L, % patients with (p = 0.019) 0.6 ± 0.61 vs. HbA1c at 0.18 ± 0.84 vs. (1) Fasting and glucose/glucagon, at least one episode (3) No episodes 0.9 ± 0.97 baseline 0.15 ± 0.63 pre-meal BG (3) RHI \ aspart = lispro mean events per (aspart vs. (3) DKA (p = NS) patient per 30 days (2) Mean change (2) Post-dinner (p = 0.002) lispro p = NS, (± SD) from baseline BG aspart vs. RHI to endpoint (2) Nocturnal hypo: p = 0.004) (3) BG at minor or major 02:00 h (3) 3.7 ± 3.6 vs. occurring between 4.4 ± 4.7 vs. midnight and 4.8 ± 4.2 06:00 h (p = NS) (3) Minor hypo: (4) 6.7 ± 5.4 vs. symptomatic and 10.5 ± 8.1 vs. BG \ 2.8 mmol/L 10.5 ± 8.9 (4) Minor hypo: (aspart vs. symptoms with or lispro without low BG p = 0.044, aspart vs. RHI p = 0.034) Diabetes Ther (2018) 9:891–917 913 Table 3 continued First author, Analog HbA1c outcomes BG levels Hypoglycemia Secondary outcomes year (n) vs. Definition Results Definition Results (mmol/L) Definition Results Measures Results comparator (n) van Bon, Glulisine Tested at (1) 7.31 ± 0.71 vs. 7-point SMBG (1) glulisine = Self-reported (1) 1.63 ± 4.50 (1) Infusion-set (1) No difference between groups 2011 [40] (256) vs. baseline and 7.33 ± 0.71 vs. 1 day before aspart = RHI vs. occlusion (1) Severe (2) 1 vs. 0 vs. 0 aspart endpoint 7.28 ± 0.71 endpoint 1.38 ± 7.73 (2) (glulisine vs. aspart) 9.2 hypoglycemia: (2) DKA (256) vs. (p = NS) vs. (3) 0.14 ± 0.43 vs. 0.06 ± 0.22 (1) Mean (1) Fasting and vs. 8.6 (p \ 0.021) BG \ 2.8 mmol/L lispro 1.06 ± 3.81 (3) Hyperketonemia, vs. 0.06 ± 0.18 (glulisine vs. (± SD) (2) 7.32 ± 0.73 vs. pre-meal and requiring (256) (3) (glulisine vs. lispro) 8.8 rate per month aspart HbA1c at 7.26 ± 0.76 vs. SMBG assistance or IV (p = NS) vs. 8.2 (p \ 0.018) (± SD) baseline 7.31 ± 0.74 glucose/glucagon, p = 0.01, glulisine vs. lispro (2) Mean post- (2) 0.61 ± 2.20 (p = NS) episodes per patient- p = 0.02) (2) Mean lunch BG vs. 0.33 ± 1.18 year (± SD) HbA1c at (3) 28 vs. 31 vs. 30 vs. 0.36 ± 1.40 (3) Mean endpoint (p = NS) (2) Nocturnal severe (glulisine vs. nocturnal hypo: severe hypo aspart (3) Percentage of BG occurring at night patients with p = 0.044, HbA1c \ 7.0 (3) Symptomatic hypo: glulisine vs. at endpoint, % BG \ 3.9 mmol/L lispro and symptomatic p = 0.070) (4) Nocturnal (3) 73.84 ± symptomatic hypo: 82.10 vs. symptomatic hypo 65.01 ± 72.07 occurring at night vs. 62.69 ± 72.87 (glulisine vs. aspart p = 0.008, glulisine vs. lispro p \ 0.001) (4) 12.80 ± 18.02 vs. 9.66 ± 13.75 vs. 9.48 ± 13.28 (glulisine vs. aspart p \ 0.001, glulisine vs. lispro p \ 0.001) BMI Body mass index, CI confidence interval, CSII continuous subcutaneous insulin infusion, DKA diabetic ketoacidosis, FFA free fatty acids, IV intravenous, PG plasma glucose, SE standard error, Aspart insulin aspart, BG blood glucose, BMI body mass index, BW body weight, CI confidence interval, CSII continuous subcutaneous insulin infusion, DKA diabetic ketoacidosis, DTSQ Diabetes Treatment Satisfaction Questionnaire, FFA free fatty acids, glulisine insulin glulisine, IV intravenous, hypo hypoglycemia, lispro insulin lispro, NR not reported, NS not statistically significant, PG plasma glucose, RHI regular human insulin, SMBG self-measured blood glucose 914 Diabetes Ther (2018) 9:891–917 (95% CI - 1.21; 0.15), a mean difference for analysis was possible, primarily in the CSII PPG of – 1.63 mmol/L (95% CI - 1.71; –1.54), a population. Those results indicate that RAIAs in risk difference for severe hypoglycemic episodes CSII lower post-breakfast BG and possibly of - 0.01 (95% CI - 0.04; 0.02) and a mean HbA1c to a greater extent than RHI, without an difference in rate of any hypoglycemic episode increased risk of hypoglycemia. Although many of - 0.75 (95% CI - 2.21; 0.72). The mean dif- trials have been published using CSII versus ference in HbA1c was - 0.19% (95% CI - 0.46; multiple daily injections (MDI) in pediatric 0.08); - 2.1 mmol/mol (95% CI - 5.0; 0.9) with subjects, we did not review those here because it RAIAs compared to RHI after 3 or 4 months of was impossible to separate the effects of the treatment. RAIA from those of the mode of treatment (CSII or MDI). DISCUSSION CONCLUSIONS This systematic review and meta-analysis sum- marizes the safety and efficacy of RAIAs in Rapid-acting insulin analogs appear to be safe populations of patients who are either typically and effective in these three special populations excluded from clinical trials (i.e. due to preg- of people with T1D. However, additional trials nancy) or require dedicated trials (i.e. children would be helpful, and head-to-head trials would and adolescents, patients using CSII). Overall, be necessary to detect any statistical differences for insulin lispro and insulin aspart, data across among them, should they exist. The lack of all three special populations indicate that their clinically relevant differences in performance safety and efficacy are comparable with, and in among the RAIAs make other factors, such as some cases significantly better than, RHI. Data cost, availability and patient/provider prefer- also suggest, from the more limited results ence, more important. Finally, studies have now available, similar characteristics for insulin been published indicating the PK advantages of glulisine. a new formulation of insulin aspart in clinical There are no head-to-head RCTs individually development (faster aspart) over conventional comparing all three RAIAs with each other in insulin aspart [46], including in pediatric pop- the pediatric CSII setting. However, in 2009, the ulations [47] and in those using CSII [48]. Institute for Quality and Efficiency in Health Studies addressing the performance of new fast- Care in Germany, acknowledging the limited acting insulin aspart versus insulin glulisine and amount of data at the time, concluded that insulin lispro in these special populations in a there were no significant differences between clinical setting would be advantageous. RHI and any of the three RAIAs in terms of key efficacy and safety endpoints [44]. The UK National Institute for Health and Care Excel- ACKNOWLEDGEMENTS lence (NICE) guidelines specifically indicate that RAIAs are preferred over RHI for use in CSII for pediatric patients [45]. Funding. The authors received no funding Given the lack of any major safety concerns for data abstraction, meta-analysis and writing. when the three RAIAs were studied individually, This work was supported by Novo Nordisk A/S. there is no a priori reason to suspect that head- Novo Nordisk A/S funded the writing and edit- to-head trials would reveal any substantive dif- ing assistance as well as the article publication ferences in safety among them in special pop- charge, and reviewed the manuscript for scien- ulations. However, without randomized tific accuracy. The authors determined which comparative trials in these special populations, studies were eligible for the review and meta- that conclusion remains speculative. Due to the analysis, performed the meta-analysis and made limited number of studies and the heterogene- the decision to submit the manuscript for ity of the outcome measures, only limited meta- publication. Diabetes Ther (2018) 9:891–917 915 Medical Writing and Editorial Assis- provide a link to the Creative Commons license, tance. The authors are grateful to Gary and indicate if changes were made. Patronek, Helen Marshall and Erin Slobodian, of Watermeadow Medical, an Ashfield company, part of UDG Healthcare plc, for writing and REFERENCES editing assistance in the development of this manuscript. Gary Patronek also performed the literature search and assisted with data abstrac- 1. American Diabetes Association. Standards of medi- cal care in diabetes—2017. Diabetes Care. tion. This assistance was funded by Novo Nor- 2017;40[Suppl 1]:S1–135. disk A/S, who also had a role in the review of the manuscript for scientific accuracy. 2. Eli Lilly. Humalog prescribing information. www. accessdata.fda.gov/drugsatfda_docs/label/2013/020 563s115lbl.pdf; 2013. Accessed 29 Jan 2018. Authorship. All named authors meet the International Committee of Medical Journal 3. Novo Nordisk. NovoLog prescribing information. Editors (ICMJE) criteria for authorship for this www.accessdata.fda.gov/drugsatfda_docs/label/2015/ article, take responsibility for the integrity of 020986s082lbl.pdf; 2015. Accessed 29 Jan 2018. the work as a whole, and have given their 4. Sanofi Aventis. Apidra prescribing information. approval for this version to be published. www.accessdata.fda.gov/drugsatfda_docs/label/2015/ 021629s030lbl.pdf; 2015. Accessed 29 Jan 2018. Disclosures. Kirsten Nørgaard serves as adviser to Medtronic, Abbott and Novo Nordisk, 5. Eli Lilly. Humalog product information. www.ema. europa.eu/ema/index.jsp?curl=pages/medicines/hum owns shares in Novo Nordisk, has received an/medicines/000088/human_med_000820.jsp& research grants from Roche and Novo Nordisk mid=WC0b01ac058001d124; 2017. Accessed 29 and has received fees for speaking from Med- Jan 2018. tronic, Roche, Rubin Medical, Sanofi, Novo 6. Novo Nordisk. NovoRapid product information. Nordisk and Bayer. Ponnusamy Saravanan has www.ema.europa.eu/ema/index.jsp?curl=pages/ received honoraria for serving on advisory medicines/human/medicines/000258/human_med_ boards and speaker fees for all three rapid-acting 000935.jsp&mid=WC0b01ac058001d124; 2017. insulin analog-producing companies (Novo Accessed 29 Jan 2018. Nordisk, Sanofi and Lilly). Nithya Sukumar and 7. Sanofi Aventis. Apidra product information. www. Snorri B. Rafnsson declare they have no conflict ema.europa.eu/ema/index.jsp?curl=pages/medicines/ of interest. human/medicines/000557/human_med_000648.jsp& mid=WC0b01ac058001d124; 2017. Accessed 29 Jan Compliance with Ethics Guidelines. This article is based on previously conducted studies, 8. Bohn B, Karges B, Vogel C, et al. 20 years of pedi- and does not involve any new studies of human atric benchmarking in Germany and Austria: age- or animal subjects performed by any of the dependent analysis of longitudinal follow-up in 63,967 children and adolescents with type 1 dia- authors. betes. PLoS One. 2016;11:e0160971. Data Availability. Data used in the meta- 9. Mooradian AD. Special considerations with insulin analysis are included in Fig. 1. therapy in older adults with diabetes mellitus. Drugs Aging. 2011;28:429–38. Open Access. This article is distributed 10. Kerr D, Wizemann E, Senstius J, Zacho M, Ampudia- under the terms of the Creative Commons Blasco FJ. Stability and performance of rapid-acting Attribution-NonCommercial 4.0 International insulin analogs used for continuous subcutaneous License (http://creativecommons.org/licenses/ insulin infusion: a systematic review. J Diabetes Sci by-nc/4.0/), which permits any non- Technol. 2013;7:1595–606. commercial use, distribution, and reproduction 11. Home PD. The pharmacokinetics and pharmaco- in any medium, provided you give appropriate dynamics of rapid-acting insulin analogues and credit to the original author(s) and the source, 916 Diabetes Ther (2018) 9:891–917 their clinical consequences. Diabetes Obes Metab. young children with type 1 diabetes. Diabetes Care. 2012;14:780–8. 2006;29:2311–2. 12. Chiang JL, Kirkman MS, Laffel LM, Peters AL, Type 23. Danne T, Ra ˚ stam J, Odendahl R, et al. Parental 1 Diabetes Sourcebook Authors. Type 1 diabetes preference of prandial insulin aspart compared with through the life span: a position statement of the preprandial human insulin in a basal-bolus American Diabetes Association. Diabetes Care. scheme with NPH insulin in a 12-wk crossover 2014;37:2034–54. study of preschool children with type 1 diabetes. Pediatr Diabetes. 2007;8:278–85. 13. Hod M, Damm P, Kaaja R, et al. Fetal and perinatal outcomes in type 1 diabetes pregnancy: a random- 24. Deeb LC, Holcombe JH, Brunelle R, et al. Insulin ized study comparing insulin aspart with human lispro lowers postprandial glucose in prepubertal insulin in 322 subjects. Am J Obstet Gynecol. children with diabetes. Pediatrics. 2008;198(186):e1–7. 2001;108:1175–9. 14. Mathiesen ER, Kinsley B, Amiel SA, et al. Maternal 25. Fairchild JM, Ambler GR, Genoud-Lawton CH, et al. glycemic control and hypoglycemia in type 1 dia- Insulin lispro versus regular insulin in children with betic pregnancy: a randomized trial of insulin type 1 diabetes on twice daily insulin. Pediatr Dia- aspart versus human insulin in 322 pregnant betes. 2000;1:135–41. women. Diabetes Care. 2007;30:771–6. 26. Ford-Adams ME, Murphy NP, Moore EJ, et al. 15. Persson B, Swahn ML, Hjertberg R, et al. Insulin Insulin lispro: a potential role in preventing noc- lispro therapy in pregnancies complicated by type 1 turnal hypoglycaemia in young children with dia- diabetes mellitus. Diabetes Res Clin Pract. betes mellitus. Diabet Med. 2003;20:656–60. 2002;58:115–21. 27. Holcombe JH, Zalani S, Arora VK, Mast CJ, Lispro in 16. Di Cianni G, Volpe L, Ghio A, et al. Maternal Adolescents Study Group. Comparison of insulin metabolic control and perinatal outcome in women lispro with regular human insulin for the treatment with gestational diabetes mellitus treated with lis- of type 1 diabetes in adolescents. Clin Ther. pro or aspart insulin: comparison with regular 2002;24:629–38. insulin. Diabetes Care. 2007;30:e11. 28. Pan ´ kowska E, Nazim J, Szalecki M, Urban M. Equal 17. Mecacci F, Carignani L, Cioni R, et al. Maternal metabolic control but superior caregiver treatment metabolic control and perinatal outcome in women satisfaction with insulin aspart in preschool chil- with gestational diabetes treated with regular or dren. Diabetes Technol Ther. 2010;12:413–8. lispro insulin: comparison with non-diabetic preg- nant women. Eur J Obstet Gynecol Reprod Biol. 29. Philotheou A, Arslanian S, Blatniczky L, Peterkova 2003;111:19–24. V, Souhami E, Danne T. Comparable efficacy and safety of insulin glulisine and insulin lispro when 18. Pettitt DJ, Ospina P, Howard C, Zisser H, Jovanovic given as part of a basal-bolus insulin regimen in a L. Efficacy, safety and lack of immunogenicity of 26-week trial in pediatric patients with type 1 dia- insulin aspart compared with regular human insu- betes. Diabetes Technol Ther. 2011;13:327–34. lin for women with gestational diabetes mellitus. Diabet Med. 2007;24:1129–35. 30. Tupola S, Komulainen J, Ja ¨a ¨skela ¨inen J, Sipila ¨ I. Post-prandial insulin lispro vs. human regular 19. Jovanovic L, Ilic S, Pettitt DJ, et al. Metabolic and insulin in prepubertal children with type 1 diabetes immunologic effects of insulin lispro in gestational mellitus. Diabet Med. 2001;18:654–8. diabetes. Diabetes Care. 1999;22:1422–7. 31. Bode B, Weinstein R, Bell D, et al. Comparison of 20. McCance DR, Damm P, Mathiesen ER, et al. Eval- insulin aspart with buffered regular insulin and uation of insulin antibodies and placental transfer insulin lispro in continuous subcutaneous insulin of insulin aspart in pregnant women with type 1 infusion: a randomized study in type 1 diabetes. diabetes mellitus. Diabetologia. 2008;51:2141–3. Diabetes Care. 2002;25:439–44. 21. Lloyd A, Townsend C, Munro V, Twena N, Nielsen 32. Guerci B, Meyer L, Salle ´ A, et al. Comparison of S, Holman A. Cost-effectiveness of insulin aspart metabolic deterioration between insulin analog and compared to human insulin in pregnant women regular insulin after a 5-hour interruption of a with type 1 diabetes in the UK. Curr Med Res Opin. continuous subcutaneous insulin infusion in type 1 2009;25:599–605. diabetic patients. J Clin Endocrinol Metab. 1999;84:2673–8. 22. Cherubini V, Iannilli A, Iafusco D, et al. Premeal insulin treatment during basal-bolus regimen in Diabetes Ther (2018) 9:891–917 917 33. Hoogma RP, Schumicki D. Safety of insulin gluli- 42. Tubiana-Rufi N, Coutant R, Bloch J, et al. Special sine when given by continuous subcutaneous management of insulin lispro in continuous sub- infusion using an external pump in patients with cutaneous insulin infusion in young diabetic chil- type 1 diabetes. Horm Metab Res. 2006;38:429–33. dren: a randomized cross-over study. Horm Res. 2004;62:265–71. 34. Johansson UB, Adamson UC, Lins PE, Wredling RA. Improved blood glucose variability, HbA1c insu- 43. Weinzimer SA, Ternand C, Howard C, Chang CT, man Infusat and less insulin requirement in IDDM Becker DJ. Laffel LM; Insulin Aspart Pediatric Pump patients using insulin lispro in CSII. The Swedish Study Group. A randomized trial comparing con- Multicenter Lispro Insulin Study. Diabetes Metab. tinuous subcutaneous insulin infusion of insulin 2000;26:192–6. aspart versus insulin lispro in children and adoles- cents with type 1 diabetes. Diabetes Care. 35. Melki V, Renard E, Lassmann-Vague V, et al. 2008;31:210–5. Improvement of HbA1c and blood glucose stability in IDDM patients treated with lispro insulin analog 44. Institute for Quality and Efficiency in Health Care: in external pumps. Diabetes Care. 1998;21:977–82. executive summaries. Rapid-acting insulin ana- logues in children and adolescents with diabetes 36. Raskin P, Holcombe JH, Tamborlane WV, et al. A mellitus type 1-follow-up commission: executive comparison of insulin lispro and buffered regular summary of final report A08-01, version 1.0. human insulin administered via continuous sub- 2005–2009. Cologne, Germany: Institute for Qual- cutaneous insulin infusion pump. J Diabetes Com- ity and Efficiency in Health Care (IQWiG); 2009. plications. 2001;15:295–300. 45. UK National Institute for Health and Care Excel- 37. Renner R, Pfu ¨ tzner A, Trautmann M, Harzer O, lence. Diabetes (type 1 and type 2) in children and Sauter K, Landgraf R. Use of insulin lispro in con- young people: diagnosis and management. NICE tinuous subcutaneous insulin infusion treatment. guideline [NG18]. 2015. www.nice.org.uk/ Results of a multicenter trial. German Humalog- guidance/ng18 (Accessed 29 Jan 2018). CSII Study Group. Diabetes Care. 1999;22:784–8. 46. Heise T, Pieber TR, Danne T, Erichsen L, Haahr H. A 38. Schmauss S, Ko ¨ nig A, Landgraf R. Human insulin pooled analysis of clinical pharmacology trials analogue [LYS(B28), PRO(B29)]: the ideal pump investigating the pharmacokinetic and pharmaco- insulin? Diabet Med. 1998;15:247–9. dynamic characteristics of fast-acting insulin aspart in adults with type 1 diabetes. Clin Pharmacokinet. 39. Tamborlane WV, Renard E, Wadwa RP, et al. Gly- 2017;56:551–9. cemic control after 6 days of insulin pump reservoir use in type 1 diabetes: results of double-blind and 47. Fath M, Danne T, Biester T, Erichsen L, Kordonouri open-label cross-over trials of insulin lispro and O, Haahr H. Faster-acting insulin aspart provides insulin aspart. J Diabetes. 2015;7:270–8. faster onset and greater early exposure vs. insulin aspart in children and adolescents with type 1 dia- 40. van Bon AC, Bode BW, Sert-Langeron C, DeVries betes mellitus. Pediatr Diabetes. 2017;18:903–10. JH, Charpentier G. Insulin glulisine compared to insulin aspart and to insulin lispro administered by 48. Heise T, Zijlstra E, Nosek L, Rikte T, Haahr H. continuous subcutaneous insulin infusion in Pharmacological properties of faster-acting insulin patients with type 1 diabetes: a randomized con- aspart vs. insulin aspart in patients with type 1 trolled trial. Diabetes Technol Ther. diabetes receiving continuous subcutaneous insulin 2011;13:607–14. infusion: a randomized, double-blind, crossover trial. Diabetes Obes Metab. 2017;19:208–15. 41. Zinman B, Tildesley H, Chiasson JL, Tsui E, Strack T. Insulin lispro in CSII: results of a double-blind crossover study. Diabetes 1997;46:440–3. Erratum. Diabetes. 1997;46:1239. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Diabetes Therapy Springer Journals

Efficacy and Safety of Rapid-Acting Insulin Analogs in Special Populations with Type 1 Diabetes or Gestational Diabetes: Systematic Review and Meta-Analysis

Free
27 pages

Loading next page...
 
/lp/springer_journal/efficacy-and-safety-of-rapid-acting-insulin-analogs-in-special-UEX11OBPft
Publisher
Springer Journals
Copyright
Copyright © 2018 by The Author(s)
Subject
Medicine & Public Health; Internal Medicine; Diabetes; Cardiology; Endocrinology
ISSN
1869-6953
eISSN
1869-6961
D.O.I.
10.1007/s13300-018-0411-7
Publisher site
See Article on Publisher Site

Abstract

Diabetes Ther (2018) 9:891–917 https://doi.org/10.1007/s13300-018-0411-7 REVIEW Efficacy and Safety of Rapid-Acting Insulin Analogs in Special Populations with Type 1 Diabetes or Gestational Diabetes: Systematic Review and Meta-Analysis . . . Kirsten Nørgaard Nithya Sukumar Snorri B. Rafnsson Ponnusamy Saravanan Received: February 6, 2018 / Published online: April 5, 2018 The Author(s) 2018 suitable studies for review and inclusion in a ABSTRACT meta-analysis. Eligible studies were randomized controlled trials that reported data on relevant Introduction: To assess the efficacy and safety clinical outcomes. A different reviewer abstrac- of three available rapid-acting insulin analogs ted data for each of the three subpopulations, (insulins lispro, aspart and glulisine, respec- and one reviewer abstracted data for all three. tively) in pregnant women, children/adoles- Any differences were resolved by consensus or cents and people using continuous by consulting a fourth reviewer. subcutaneous insulin infusion (CSII) with type Results: In people on CSII, rapid-acting insulin 1 diabetes. analogs lowered postprandial plasma glucose Methods: PubMed, EMBASE and Cochrane post-breakfast to a greater extent than did reg- Reviews were searched electronically, and their ular human insulin (RHI) (mean difference: bibliographies examined to identify - 1.63 mmol/L [95% confidence interval - 1.71; - 1.54]), with a comparable risk of Enhanced content To view enhanced content for this hypoglycemia and a trend for lower glycated article go to https://doi.org/10.6084/m9.figshare. hemoglobin. In the pediatric population, gly- cemic control was similar with rapid-acting Electronic supplementary material The online insulin analogs and RHI, with no safety con- version of this article (https://doi.org/10.1007/s13300- cerns. Meta-analysis indicated severe hypo- 018-0411-7) contains supplementary material, which is glycemic events were comparable for rapid- available to authorized users. K. Nørgaard S. B. Rafnsson Department of Endocrinology, Copenhagen Centre for Primary Health and Social Care, London University Hospital Hvidovre, Kettega ˚ rd Alle ´ 30, Metropolitan University, 166–220 Holloway Road, 2650 Hvidovre, Denmark London N7 8DB, UK K. Nørgaard S. B. Rafnsson Steno Diabetes Center Copenhagen, Niels Department of Epidemiology and Public Health, Steensensvej 2, 2820 Gentofte, Denmark University College London, 1–19 Torrington Place, London WC1E 7HB, UK N. Sukumar  P. Saravanan (&) Diabetes, Endocrinology and Metabolism, Division P. Saravanan of Health Sciences, Warwick Medical School, Department of Diabetes, Endocrinology and University of Warwick, Coventry CV4 7AL, UK Metabolism, George Eliot Hospital, e-mail: p.saravanan@warwick.ac.uk Nuneaton CV10 7DJ, UK 892 Diabetes Ther (2018) 9:891–917 acting insulin analogs versus RHI (risk differ- in the chemical composition of their formula- ence: 0.00 [95% confidence interval - 0.01; tions [10], but their pharmacokinetic (PK) and 0.01]). In the pregnancy group, insulin lispro pharmacodynamic (PD) profiles are similar and insulin aspart were safe and effective for (Electronic Supplementary Material [ESM] both mother and fetus, with glycemic control Table S1). A large body of clinical studies indi- being at least as good as with RHI. There were cates that these three RAIAs have similar effi- no data on insulin glulisine during pregnancy. cacy and safety [11] and that they are preferred Conclusion: Rapid-acting insulin analogs over regular human insulin (RHI) for use in appear generally safe and effective in these adults with T1D due to their lower risk of special populations; however, additional trials hypoglycemia [12]. would be helpful. However, certain subgroups of patients with Funding: Novo Nordisk A/S. T1D (e.g. children and adolescents, pregnant women and people using CSII) are typically excluded from trials conducted for regulatory Keywords: CSII; Pediatrics; Pregnancy; Rapid- approval to ensure a more homogeneous group acting insulin analogs; Type 1 diabetes of patients. A consequence of excluding these patients INTRODUCTION from regulatory trials is that there is some uncertainty about the clinical profile of RAIAs Many people with type 1 diabetes (T1D) receive in patients who may have unique metabolic, developmental, cognitive or behavioral issues insulin therapy [1]. Rapid-acting insulins are typically used to control postprandial plasma that materially affect the suitability of any medication. With RAIAs in widespread clinical glucose (PPG) excursions, whereas long-acting basal insulins are used to control fasting glu- use, it is timely to examine the available evi- dence for their performance in special popula- cose. Rapid-acting insulins are used as part of a basal–bolus injection regimen as well as for tions. Thus, we performed a systematic review and meta-analysis of published data (PROSPERO continuous subcutaneous insulin infusion registration #CRD42016043006). (CSII). Three rapid-acting insulin analogs (RAIAs) are currently available in the USA and Europe: insulin lispro (Humalog ; Eli Lilly, METHODS Indianapolis, IN, USA), insulin aspart (Novolog in the USA and NovoRapid in the Sources of Data and Search Criteria EU; Novo Nordisk, Bagsværd, Denmark) and insulin glulisine (Apidra ; Sanofi Aventis, The search terms ‘‘insulin lispro’’ (MeSH Terms) Bridgewater, NJ, USA). All three of these RAIAs OR ‘‘insulin’’ (All Fields) AND ‘‘lispro’’ (All are also approved for the pediatric T1D patient Fields) OR ‘‘insulin lispro’’ (All Fields) OR ‘‘lis- population, although the ages on the product pro’’ (All Fields) OR aspart (All Fields) OR gluli- inserts for which there are data vary by product sine (All Fields) were used to search the PubMed, as well as by country (e.g. USA: insulin lispro, EMBASE and the Cochrane Reviews databases children C 3 years of age; insulin aspart, electronically on 1 June 2016 to identify records C 2 years; insulin glulisine, C 4 years [2–4]; EU: for further examination. The titles and abstracts insulin lispro, age not specified; insulin aspart, (and, when necessary, full papers) were then C 1 years; insulin glulisine, C 6 years) [5–7]. In screened to identify papers potentially report- addition, despite the concerns in some coun- ing relevant in vivo data on safety or efficacy in tries, the use of RAIAs is very high in T1D randomized controlled trials (RCTs) involving patients in general and is almost 100% among one of three special populations (children and those patients using CSII [8]. adolescents with T1D; pregnant women with The three insulin analogs lispro, aspart and pre-gestational T1D or gestational diabetes glulisine differ in how their molecular structure [GDM]; people with T1D using CSII). We did has been modified from human insulin [9] and Diabetes Ther (2018) 9:891–917 893 not set a date range and therefore included any by consensus or, if necessary, by consulting a study published up to the date of the search. fourth reviewer (PS) not involved in the original Reference lists of retrieved publications and data abstraction process for that population. All targeted review articles of RAIAs were also reviewers are qualified at a post-doctorate level. searched to identify additional records that might be provisionally relevant. Data Synthesis and Statistical Analysis Selection of Studies and Eligibility Criteria For each of the three populations, whenever possible, we extracted data on glycemic control Records identified as provisionally relevant were (HbA1c at baseline and end of trial; percentage then further examined for eligibility to verify of patients achieving HbA1c target[s] as speci- that they were indeed RCTs, either blinded or fied in individual trials; 7- or 8-point self- open-label and of parallel or crossover design, in measured blood glucose [SMBG]); and safety one of the target special populations and that endpoints (e.g. severe, nocturnal, overall hypo- they reported data on one of the identified glycemia; hyperglycemia/ketosis; fetal out- clinical outcomes of interest. Eligibility for the comes). For completeness, we also extracted three study populations was as follows: for the other secondary endpoints if available. Each of pregnancy population, women with either pre- the retrieved studies was assessed for study gestational T1D or GDM; for the pediatric pop- quality, including sample size, reporting of ulation, children or adolescents aged\ 18 years methods, reporting of results and risk of bias. with T1D; for the CSII population, people with Finally, using this information, we graded each T1D of any age using an insulin pump. study according to the following scale (-, ?, The criterion for the treatment in eligible ??), with - indicating poor quality, ? indi- studies was the administration of one of the cating average quality and ?? indicating good three RAIAs compared with either RHI or each quality. other; trials in which the effect of the RAIA All authors reviewed the available outcomes could not be isolated (as in basal–bolus trials in data for each eligible study and decided which which different basal insulins were used in each studies had efficacy and/or safety data that arm) were excluded. We did not pre-specify a would be suitable for data combination and minimum duration for the studies, although meta-analysis within each of the three special very short-term (e.g. 1- to 2-day meal tests populations. Suitability was determined by designed to study PK/PD) were excluded. Out- authors’ assessment of comparability of the come measures of interest included glycated outcomes assessed. hemoglobin (HbA1c), fasting blood glucose Meta-analysis was conducted using Review (FBG) or plasma glucose, PPG after any or all of Manager 5 (RevMan 5.3 [http://community. the three main meals, hypoglycemia, hyper- cochrane.org/tools/review-production-tools/rev glycemia, diabetic ketoacidosis and/or pump/- man-5]) software for Cochrane Reviews. Two catheter occlusion and, for pregnancy, fetal different outcome measures were used for the outcomes. safety variables in the meta-analysis (namely risk difference for severe and any hypoglycemic episodes per month) and three outcome mea- Data Extraction sures for the efficacy variables (namely mean difference in fasting and postprandial blood A standardized data abstraction form was glucose [BG] and in HbA1c). All models were adapted for use in this systematic review. A run as random-effects models. Statistical different reviewer abstracted data for each of the heterogeneity was calculated by using the I three subpopulations (KN, NS, SR) using the statistic, and publication bias was assessed by inclusion and exclusion criteria, and one using a funnel plot and the Egger’s test. reviewer abstracted data for all three of the subpopulations. Any differences were resolved 894 Diabetes Ther (2018) 9:891–917 Compliance with Ethics Statement mixture formulations; outcomes of interest were not reported; PK/PD studies in which basal insu- lins were the focus of the study; study primarily This article is based on previously conducted addressing stability in pumps; health–economics studies and does not involve any new studies of studies without efficacy or safety outcomes; paper human or animal subjects performed by any of written in a language other than English; dupli- the authors. cate record; or study not applicable to this review for other reasons. A flow diagram showing the RESULTS number of retrieved, excluded and included records is shown in Fig. 1. Prior to submission of Our search of the PubMed, EMBASE and the this manuscript (August 2017), PubMed and the Cochrane Reviews databases resulted in the Cochrane Reviews were searched as previously identification of seven eligible studies in preg- described and no additional studies that would nancy, nine in pediatric patients and 13 in people have been provisionally eligible were identified. using CSII. Records were assessed to be ineligible and thereby excluded from the systematic review Pregnancy and meta-analysis for numerous reasons, includ- ing the study not being a RCT (observational Characteristics of the seven eligible studies in studies, letters, case reports or review articles were pregnancy (two in pre-gestational T1D [13–15] excluded); ineligible population; in vitro or ani- and four in GDM [16–19], with two reporting mal studies; study dealt with one of the RAIAs in different outcomes from the same population Fig. 1 Flow diagram showing the number of retrieved, excluded and included records. The dagger () indicates that the continuous subcutaneous insulin infusion (CSII) group included two pediatric CSII studies. The asterisk (*) indicates that in the CSII meta-analyses, the number of studies included varied depending on the outcome assessed. MDI Multiple daily injections, PD pharmacodynamics, PK pharmacokinetics, RCT randomized controlled trial Diabetes Ther (2018) 9:891–917 895 Table 1 Study outcomes: pregnancy First author, Analog HbA1c outcomes BG levels Hypoglycemia Secondary outcomes year of study (n) vs. Definition Results Definition Results (mmol/L) Definition Results Measures Results comparator (n) Hod, 2008 Aspart (157) Tested in early (1) - 0.04% 8-point SMBG performed (1) 6.82 vs. 6.82 Self-reported diary (1) 0.48 (0.20, (1) Mean (± SD) (1) 87.6 ± 12.0 [13] and vs. RHI pregnancy and (- 0.18; for 1 week in each 1.14) overall treatment vs. (2) 6.96 vs. 7.10 (1) Relative risk Mathiesen, (165) end of each 0.11) trimester (p = NS) satisfaction score 83.4 ± 15.3 (95% CI) of major (3) 6.23 vs. 6.48 2007 [14] trimester (p = NS) (p = 0.03) (1) Mean 24-h BG at first nocturnal (2) 0.72 (0.36; (2) Corrected BW (4) NR, p = 0.044 hypoglycemia 1.46) (SEM) (1) Mean difference (2) - 0.08 trimester (2) 3438 (71.5) (95% CI) between (- 0.23; (p = NS) vs. 3555 (72.9) (5) NR, p = 0.153 (2) Mean 24-h BG at (2) Relative risk (3) Preterm birth aspart and RHI at 0.06) (p = 0.09) second trimester (95% CI) of major (6) NR, p = 0.0007 (4) Neonatal hypo end of second (p = NS) hypoglycemia in (3) 20.3% vs. (3) Mean 24-h BG at third trimester (%) aspart vs. RHI 30.6% trimester (2) Mean difference (p = 0.05) (4) Mean difference in at end of third (4) 33.6 vs. 39.7% post-breakfast glucose in trimester aspart vs. RHI in first trimester (5) Mean difference in post-breakfast BG in second trimester (6) Mean difference in post-breakfast BG in third trimester Persson, 2002 Lispro (16) Tested monthly (1) 6.5 Combined values in second (1) 6.50 ± 3.18 vs. Self-reported diary (1) 0 vs. 2 (1) Progression of (1) 18.8 vs. 35.2 [15] vs. RHI (4.8–8.6) and third trimester of 8.56 ± 3.55 retinopathy, % in (1) Median (range) (1) Number of (2) 5.5 vs. 3.9 (2) No difference (17) vs. 6.6 6-point SMBG (p \ 0.01) each group HbA1c at baseline patients (p \ 0.05) in perinatal (4.5–8.6) (6–8 weeks (1) Mean (± SD) post- (2) 0.40 ± 3.20 vs. experiencing severe (2) Neonatal outcomes outcome or gestation) (%) (2) 5.4 breakfast BG (mmol/L) 1.81 ± 3.42 hypo (i.e., anthropometry, neonatal (4.3–5.9) vs. (p \ 0.01) complications, complications (2) Median HbA1c (2) Mean post-breakfast (2) Rate of 5.3 malformations, at 24 weeks glucose increase (3) 0.70 ± 3.16 vs. biochemical hypo (4.7–6.7), symptomatic hypo) gestation 0.60 ± 3.34 (BG \ 3.0) (%) (3) Mean post-lunch p =NS (p = NS) (3) Median HbA1c glucose increase (3) 5.2 before delivery (4) - 0.24 ± 3.10 (4) Mean post-dinner (4.6–5.9) vs. vs. 0.28 ± 2.94 5.0 glucose increase (p \ 0.04) (4.6–6.7), p =NS 896 Diabetes Ther (2018) 9:891–917 Table 1 continued First author, Analog HbA1c outcomes BG levels Hypoglycemia Secondary outcomes year of study (n) vs. Definition Results Definition Results (mmol/L) Definition Results Measures Results comparator (n) Mecacci, 2003 Lispro (25) Tested at diagnosis of (1) - 0.3 ± 9-point SMBG performed (1) post-BF: NR NR (1) Neonates with (1) 12% vs. 37.5% [17] vs. RHI GDM and 0.3 vs. weekly from diagnosis to 5.93 ± 0.62 vs. cranial-thoracic (p \ 0.05) (24) delivery - 0.3 ± 38 weeks gestation 6.84 ± 0.7 circumference (CC/ (2) No difference 0.1 (p \ 0.002) CT) ratio 10-25th (1) Mean difference (1) Mean (± SD) 1-h post- between groups (p = NS) centile, % (± SD) from pre- breakfast BG (2) post-lunch: intervention to 5.91 ± 0.66 vs. (b) Neonatal (2) Mean BG 1-h post- 6.63 ± 0.88 delivery (%) anthropometry (BW, lunch BG ponderal index) and (p \ 0.01) (3) Mean BG 1-h post- complications evening meal BG (3) post-dinner: 6.21 ± 0.49 vs. (4) Mean total BG 1-h 6.71 ± 0.62 postprandial BG (p \ 0.05) (5) Mean total preprandial BG (4) total: 6.02 ± 0.59 vs. (6) Mean total 2-h 6.72 ± 0.73 postprandial BG (p \ 0.01) (5) 4.08 ± 0.45 vs. 4.13 ± 0.73 (p = NS) (6) 5.20 ± 0.62 vs. 5.44 ± 0.69 (p = NS) Di Cianni, Aspart (31) NR NR 5-point SMBG (1) 6.75 ± 1.12 vs. Number of patients 0 vs. 0 vs. 0 (1) Neonatal (1) Higher BW 2007 [16] vs. lispro 6.6 ± 1.05 vs. experiencing any anthropometry: in RHI than (1) Mean (± SD) 1-h post- (33) vs. 7.5 ± 1.3 hypo BW, CC/CT ratio aspart/lispro breakfast BG RHI (32) groups (p \ 0.05) (2) Macrosomia (2) Macrosomia: 9.6%, 12.1%, 15.6% (p = NS) CC/CT ratio lower in RHI than other two groups (p = 0.03) Diabetes Ther (2018) 9:891–917 897 Table 1 continued First author, Analog HbA1c outcomes BG levels Hypoglycemia Secondary outcomes year of study (n) vs. Definition Results Definition Results (mmol/L) Definition Results Measures Results comparator (n) Pettitt, 2007 Aspart (14) Tested at diagnosis of (1) 5.1 ± 0.4 Measured on standardized (1) 4.2 ± 0.57 vs. Self-reported diary (1) 71 (53) vs. (1) Insulin-specific (1) Low insulin- [18] vs. RHI GDM and vs. mixed meal test after 4.8 ± 0.86 69 (23) antibodies specific (1) % of subjects (13) delivery 5.3 ± 0.3 6 weeks antibodies in (2) - 1.09 ± (number of events) (2) 79 (52) vs. (2) Cross-reactive both groups (1) Diagnosis (2) 5.4 (1) Time-adjusted mean 0.55 vs. experiencing 39 (9) antibody binding (± SD) glucose - 0.54 ± 0.74 symptomatic hypo (± SD) at 36/40 (2) 1.4 ± 3.0 vs. (2) Delivery (both (p = 0.003) and 6/52 post- 1.5 ± 2.3 and groups combined) (2) Time-adjusted mean (2) % of subjects partum (%) 2.3 ± 5.4 vs. change in glucose from (3) 4.7 ± 0.19 vs. (number of events) 6.5 ? 13.7 baseline (no insulin) 5.1 ± 0.23 experiencing minor (3) Mean neonatal (p = 0.03) hypo birth weight (± SD) (3) 3.1 (0.5) vs. (3) Mean glucose at 30 min (kg) 3.0 (0.5) time point of meal test (4) 5.4 ± 0.21 vs. (p = NS) 6.2 ± 0.33 (4) Mean birth length (4) Mean glucose at 60 min (p \ 0.01) (± SD) (cm) (4) 49 (2.3) vs. 48 time point of meal test (2.4) (p = NS) Jovanovic, Lispro (19) Tested at baseline Mean 6-point SMBG (1) 5.5 ± 0.30 vs. Self-reported diary (1) 0.65 ± 0.13 (1) Change in insulin- (1) No difference 1999 [19] vs. RHI and 6 weeks after (± SEM) 7.3 ± 0.4 and meter check vs. specific and cross- in maternal (1) Mean (± SEM) post- (23) randomization HbA1c at: (p = NS) 0.93 ± 1.04 reactive response antibody breakfast hyperglycemia (1) Mean (± SEM) (p \ 0.05) from baseline to response (1) Mean (± SEM) (1) 5.47 ± rate (individual patient % (2) 4.5 ± 1.1 vs. pre-breakfast hypo delivery at baseline (%) 0.09 vs. 5.24 of all 6.8 ± 0.86 rate (individual (2) 0.78 ± 0.37 (2) No difference vs. 0.09 readings C 6.67 mmol/ (p = NS) patient % of all vs. (2) Neonatal in neonatal (2) Mean at 6 weeks (p = NS) L), (%) readings \ 3.1 1.98 ± 0.81 anthropometry anthropometry (3) 2.0 ± 0.51 vs. (3) Mean change mmol/L) (%) and outcomes (2) 5.12 ± (2) Mean post-lunch 2.6 ± 0.54 (3) 1.26 ± 0.43 (3) Neonatal glucose from baseline to 6 0.11 vs. hyperglycemia rate (p = NS) (2) Mean pre-lunch vs. (3) No neonatal weeks (% 5.16 ± 0.12 hypo rate 1.43 ± 0.86 hypo- or reduction) (3) Mean post-evening (4) 4.0 ± 0.49 vs. hyperglycemia (3) - 0.35 meal hyperglycemia rate 5.5 ± 0.47 (3) Mean post- (4) 0.88 ± 0.25 (5.7%) vs. (p \ 0.05) evening meal hypo vs. (4) Mean total postprandial - 0.07 rate 2.20 ± 0.86 hyperglycemia rate (2.8%) (4) Mean total (p = 0.002) preprandial hypo rate Aspart Insulin aspart, BF breakfast, BG blood glucose, BW body weight, CC/CT ratio of the cardiac circumference to the thoracic circumference, CI confidence interval, hypo hypoglycemia, GDM gestational diabetes, glulisine insulin glulisine, HbA1c glycated hemoglobin, lispro insulin lispro, NR not reported, NS not significant, RHI regular human insulin, SD standard deviation, SEM standard error of the mean, SMBG self-measured blood glucose 898 Diabetes Ther (2018) 9:891–917 and therefore described in a single row [13, 14]), insulin aspart (rate ratio 0.72 [95% CI 0.36; are summarized in ESM Table S2, and the results 1.46]). of those studies are shown in Table 1. Most (4/6 In a publication reporting additional data on trials) involved \ 50 participants. Study quality pregnancy outcomes by Mathiesen et al. [14], assessment is shown in ESM Table S3. Three Hod et al. [13] indicated that preterm delivery studies were graded as -, two were graded occurred in 20.3 and 30.6% of pregnancies in as ? and one was graded as ??. women receiving insulin aspart and RHI, respectively (p = 0.053). Other secondary pub- lications from these trials indicated that there Women with Pre-Existing T1D were 137 and 131 live births, 14 and 21 fetal Persson et al. compared treatment with insulin losses, and six and nine congenital malforma- lispro with RHI treatment in a group of 33 tions in these groups of women on insulin pregnant women with T1D and found that the aspart and RHI, respectively. Furthermore, BG level was significantly lower after breakfast maternal and cord blood antibody levels for (but not after other meals) with insulin lispro both RHI and insulin aspart remained low for (0.40 ± 3.20 vs. 1.81 ± 3.42 mmol/L; p \ 0.01) both treatments and were similar at 36 weeks [15] (Table 1). The 95% confidence interval (CI) gestation for the 97 women who participated in was not significantly different between treat- the substudy [20]. In a secondary analysis of ments, either at 24 weeks gestation or before data from the same trial, Lloyd et al. reported delivery. The incidence of severe hypoglycemia that these benefits were attained without was low (zero and two episodes for insulin lispro increasing the cost of treatment compared to and RHI, respectively) and there were no dif- RHI [21]. ferences in perinatal outcomes or neonatal complications. The largest trial was an international, paral- Women with GDM lel-group trial enrolling 322 women with T1D The characteristics of four RCTs using RAIAs in who at enrollment were pregnant GDM are summarized in ESM Table S2 [16–19]. for B 10 weeks or planning to become preg- The results indicate that insulin lispro was at nant. Women were randomized to either insu- least as effective as RHI and sometimes lin aspart or RHI, both in combination with demonstrated improved glycemic control (ESM neutral protamine Hagedorn insulin as the basal Table S2) [16, 17, 19]. Insulin aspart was asso- insulin [14]. HbA1c levels were comparable in ciated with significantly lower post-meal BG the two groups at the end of the second and compared with RHI [16, 18]. In the single trial third trimesters (treatment difference, insulin involving a head-to-head comparison of insulin aspart–RHI: - 0.04% [95% CI - 0.18; 0.11], aspart and insulin lispro, mean 1-h post-break- - 0.4 mmol/mol [95% CI - 2.0; 1.2]; and fast BG was similar for the two products - 0.08% [95% CI - 0.23; 0.06], - 0.9 mmol/- (6.75 ± 1.12 vs. 6.6 ± 1.05 mmol/L, respec- mol [95% CI - 2.5; 0.7], respectively). Mean tively) [16]. plasma glucose levels at 90 min post-breakfast In Mecacci et al. [17], hypoglycemia was not were significantly lower in those women reported and, in another trial, there were no receiving insulin aspart arm than in those hypoglycemic events reported for insulin receiving RHI (p = 0.044 and p = 0.001 for end aspart, insulin lispro or RHI [16]. In a study of of first and third trimesters, respectively). The women (n = 27) using insulin aspart or RHI, the mean PPG increment across all meals was lower reported percentage of participants experienc- for the insulin aspart arm than for the RHI arm ing symptomatic hypoglycemic events was at the end of the first and third trimesters (es- similar for both treatments (71 vs. 69%), but timated treatment difference: –0.75 [95% CI more participants using insulin aspart reported - 1.25; - 0.25], p = 0.003 and - 0.40 [95% CI minor hypoglycemia (79 vs. 39%) [18]. In the - 0.80; - 0.01], p = 0.044, respectively). The latter case, this was largely due to two partici- risk of major hypoglycemic events was numer- pants being prone to hypoglycemia. Neonatal ically lower, but not significantly different, for Diabetes Ther (2018) 9:891–917 899 outcomes (weight, length, physical exam) were insulin lispro, respectively; p = 0.039). The fre- good for both insulin aspart and RHI. quencies of hypoglycemic events and other adverse events were similar. Meta-Analysis The pregnancy studies were deemed to be too Meta-Analysis heterogeneous or to lack relevant information Data on glycemic control were deemed to be too for meta-analysis of either efficacy or safety heterogeneous or to lack relevant information outcomes for any of the three RAIAs. for meta-analysis, but the data on severe hypo- glycemic outcomes were able to be combined from five studies (Fig. 2a). Overall, the number Children and Adolescents of severe hypoglycemic events was low (total of 14), and a random-effects model using generic There were nine eligible studies involving inverse variance showed no difference in risk of pediatric patients with T1D [22–30]. The char- severe hypoglycemia with insulin analog treat- acteristics of these studies are presented in ESM ment, risk difference 0 (95% CI - 0.01; 0.01). Table S4 and the results are shown in Table 2. Funnel plots suggested that there was no pub- Three trials compared insulin aspart with RHI lication bias (data not shown). [22, 23, 28], five compared insulin lispro with RHI [24–27, 30] and one compared insulin lis- Patients Treated with CSII pro with insulin glulisine [29]. Most (5/9; 55.5%) trials involved \ 50 participants. Study quality assessment is shown in ESM Table S5. The largest number of eligible studies (n = 13) Four studies were graded as -, two were graded was identified for people using CSII [31–43], of as ? and three were graded as ??. Trials using which two were carried out in pediatric popu- RAIAs in CSII in children are discussed in the lations [42, 43] (ESM Table S6). The results are section ‘‘Patients Treated with CSII’’. presented in Table 3, and the study quality Overall, glycemic control (either HbA1c or assessment is shown in ESM Table S7. Two PPG) with insulin lispro or insulin aspart was studies were graded as -, three were graded equivalent to or better than that with RHI. This as ? and seven were graded as ??. One of the was also true with respect to incidence of studies consisted of two substudies, one of hypoglycemic episodes or other adverse events. which was graded as ?, and the second was There were no head-to-head trials comparing all graded as ?? due to the double-blind three RAIAs in pediatric participants. However, component. a large (n = 572), open-label, parallel-group, Eight studies compared an RAIA with RHI non-inferiority trial compared insulin glulisine and all involved insulin lispro [32, 34–38, 41], with insulin lispro [29]. Insulin glulisine was with one being a pediatric trial [42]. All were demonstrated to be non-inferior to insulin lis- crossover trials of 1–4 months’ duration. All pro (treatment difference in HbA1c: - 0.06% indicated that insulin lispro was associated with [95% CI - 0.24; 0.12]; - 0.7 [95% CI - 2.6; improved glycemic control (HbA1c) and an 1.33] mmol/mol). More children achieved incidence of hypoglycemic events that was American Diabetes Association (ADA) age- similar to or lower than RHI. specific HbA1c targets (at the time of the Three studies involved head-to-head com- study: \ 6 years, HbA1c [ 7.5 to%\ 8.5% [[ 58 parisons of insulin lispro versus insulin aspart to \ 69 mmol/mol]; 6–12 years, HbA1c \ 8.0% [33, 39, 43], with an additional trial also com- [\ 64 mmol/mol]; 13–17 years, HbA1c \ 7.5% paring RHI [31]. The largest of the three trials [\ 58 mmol/mol]; currently, the ADA recom- was a 16-week, open-label RCT in 298 subjects mends HbA1c \ 7.5% [\ 58 mmol/mol] across with T1D aged 4–18 years [43] (ESM Table S6). all pediatric age groups [1]) with insulin gluli- At 16 weeks, the HbA1c in subjects receiving sine than with insulin lispro (overall popula- insulin aspart was deemed to be non-inferior to tion: 38.4 vs. 32.0% for insulin glulisine and the HbA1c in those receiving insulin lispro, and 900 Diabetes Ther (2018) 9:891–917 Table 2 Study outcomes: pediatrics First Analog HbA1c outcomes BG levels Hypoglycemia Secondary outcomes Author, (n) vs. Definition Results Definition Results (mmol/L) Definition Results Measures Results year comparator (n) Philotheou, Glulisine Tested at (1) ? 0.10 ± 0.08 3-point SMBG at (1) 8.77 ± 0.21 vs. Reported from month 4 to (1) 0.06 ± 0.24 vs. Mean (± SD) 2.53 ± 0.68 vs. 2011 (275) vs. baseline and vs. ? endpoint 9.46 ± 0.21 treatment end 0.07 ± 0.27 increase total 4.91 ± 0.65 [29] lispro endpoint 0.16 ± 0.07 (p = 0.014) daily dose of (p = 0.007) (1) Adjusted (1) Severe hypo (± SD): (2) 0.21 ± 0.50 vs. (295) insulin from (1) Adjusted (2) - 0.06 mean (2) 9.76 ± 0.24 vs. BG \ 2.0 mmol/L and 0.20 ± 0.80 baseline, mean change (- 0.24; 0.12) (± SEM) of 9.80 ± 0.23 third-party assistance (3) 3.10 ± 4.33 vs. units/day (± SD) from pre-breakfast (p = 0.894) required or prompt (3) 38.4 vs. 32.0 2.91 ± 4.35 baseline to BG (p value for recovery following glucose (p = 0.0039) (3) 9.20 ± 0.22 vs. endpoint difference treatment, number of 9.04 ± 0.21 between episodes per patient- (2) Difference (p = 0.564) groups) month between treatments in (2) Adjusted (2) Nocturnal hypo adjusted means mean (3) Any hypo (95% CI) for (± SEM) of difference pre-main meal from baseline BG to endpoint (3) Adjusted (3) % achieving mean ADA age- (± SEM) of specific HbA1c 2-h post main targets at meal endpoint Pan´kowska, Aspart (20) Tested at (1) 7.4 ± 0.9 vs. 24-h glycemic (1) 219.8 (12.8) vs. Self-reported by parent (1) 0.1 vs. 0.0 Treatment 5.6 vs. 4.4 2010 vs. RHI baseline, 7.6 ± 1.1 control 211.8 (10.9) during CGM period satisfaction: SD (p = 0.04) (1/20) vs. (0/21) [28] (21) midpoint and measured (p = 0.55) values for mean (2) 7.6 ± 0.9 vs. (1) Severe hypo: endpoint using CGMS (2) 18 vs. 20 change in 7.6 ± 1.0 (2) 4.0 vs. 4.0 BG \ 2.8 mmol/L for 72 h at treatment (1) Mean (p = NS) accompanied by CNS 18/20 vs. 19/21 endpoint satisfaction HbA1c symptoms requiring (3) 1.1 vs. 1.4 score at end of (± SD) at (1) Mean external help, number of study 13 weeks after (± SD) of area episodes per patient-year treatment under glucose of exposure curve, mmol/ (2) Mean (2) Minor hypoglycemia: h/L HbA1c at BG \ 2.8 mmol/L that 26 weeks after (2) Difference were asymptomatic or treatment between self-treatable maximum and (3) Symptoms-only minimum hypoglycemia glucose levels over 24 h Diabetes Ther (2018) 9:891–917 901 Table 2 continued First Analog HbA1c outcomes BG levels Hypoglycemia Secondary outcomes Author, (n) vs. Definition Results Definition Results (mmol/L) Definition Results Measures Results year comparator (n) Cherubini, Aspart (NR) Tested at (1) 7.5 ± 0.8 to 7-point SMBG (1) 7.38 ± 1.0 vs. Self-reported (1) 0.045 vs. 0.035 NR NR 2006 vs. RHI baseline and 7.0 ± 0.4 vs. during study 7.94 ± 1.06 (p = 0.209) (1) Severe hypo: [22] (NR) 6-weekly 7.5 ± 1.4 to period (p = 0.175) BG \ 2.8 mmol/L, (2) 0.214 vs. 0.18 during study 7.4 ± 0.5 (1) Mean 2-h (2) number of episodes per (p = 0.117) (p = 0.018) (1) Change in postprandial Analog \ RHI, patient per day HbA1c BG p = 0.012 (2) Any hypo: (± SD) over (2) Mean fasting (3) 8.89 ± 2.89 vs. BG \ 3.9 mmol/L the study BG 9.17 ± 2.72 period (p = 0.113) (3) Mean afternoon BG (4) Analog = RHI, at endpoint p [ 0.237 (4) Decrease in mean daily BG variability from baseline Fairchild, Lispro (35) Tested at (1) 8.33 ± 0.89 7-point SMBG (1) 10.57 ± 0.26 Self-reported and BG meter (1) 0.032 vs. 0.065 Treatment 28 (80) vs. 7 (20) 2000 vs. RHI baseline and vs. 8.14 ± 0.77 done weekly vs. 9.02 ± 0.46 analysis (p = NS) satisfaction [25] (35) 6-weekly (p = NS) or 2-weekly (p = 0.001) questionnaire (1) Severe hypos: hypo 1/35 vs. 2/35 during study (1) Mean BG (2) 2.35 associated with Number (%) (2) 13.47 vs. 10.77 (1) Mean (± SD) at (-3.98; - 0.72) convulsion or coma preferring this (p = NS) HbA1c 03:00 h type of insulin p = 0.01 (2) Total recorded hypos, (± SD) at (3) 5.69 vs. 3.31, over the other (2) Mean number of episodes per endpoint difference 2.4 ± 5.1 one difference patient per 3 months (p = 0.02) (95% CI) in (3) Total recorded hypos 03:00 h BG (4) analog = comparator from 06:00 to 12:00 h between groups (p = NS) (4) Hypos with BG \ 3.0 mmol/L 902 Diabetes Ther (2018) 9:891–917 Table 2 continued First Analog HbA1c outcomes BG levels Hypoglycemia Secondary outcomes Author, (n) vs. Definition Results Definition Results (mmol/L) Definition Results Measures Results year comparator (n) Ford- Lispro (23) Tested at (1) 8.9 ± 0.3 vs. Overnight (1) 6.5 ± 1.0 vs. Self-reported (1) 2/23 vs. 1/23 Mean (range) total 0.97 (0.68–1.26) Adams, vs. RHI baseline and 8.4 ± 0.3 metabolic 7.1 ± 1.1 daily insulin vs. 0.96 (1) Severe hypo: (2) 1.6 ± 0.3 vs. 2003 (23) 4-monthly study (p = 0.5) dose, units/kg (0.53–1.22) (p = 0.14) convulsions or requiring 1.7 ± 0.3 (p = 0.2) [26] during study (p = 0.2) (1) Mean BG (2) 6.1 ± 0.8 vs. glucagon, number of (2) 8.5 ± 0.2 vs. (3) 9 vs. 6 (p = 0.06) (1) Mean (± SD) at start 6.3 ± 0.9 episodes 8.8 ± 0.3 HbA1c of overnight (p = 0.8) (4) 8 vs. 13 (p = 0.01) (2) Symptomatic (± SD) at (p = 0.47) profile (3) 138 ± 12 vs. hypoglycemia, number of (5) 27 vs. 22 crossover (2) Mean fasting 170 ± 13 episodes per patient per (p = 0.11) (2) Mean BG (p = 0.03) week (± SD) HbA1c at (3) AUC of BG (4) 158 ± 13 vs. Overnight metabolic profile endpoint (± SD) from 145 ± 12 (3) Prevalence of low BG post-EM to (p = 0.3) from post-EM to bedtime, 22:00 h: mmol/min/L BG \ 3.5 mmol/L, % (4) AUC of BG (4) Prevalence of low BG from 22:00 to from 22:00 to 04:00 h 04:00 h (5) Prevalence of low BG from 04:00 to 07:00 h Holcombe, Lispro (457) Tested at (1) 8.41 ± 1.4 vs. 8-point BG (1) 9.7 ± 3.9 vs. Self-reported (1) 5 vs. 5 (1) Mean (± SD) (1) 1.08 ± 0.32 2002 vs. RHI baseline and 8.80 ± 1.5 profiles on 8.8 ± 3.7 total daily vs. 1.05 ± 0.30 (1) Severe hypo: needing 5/457 vs. 5/547 [27] (457) 2-monthly 2 days at (p \ 0.001) insulin dose, (p \ 0.001) (p = NS) third-party assistance or during study baseline and (2) 1.0 ± 1.9 vs. units/kg (2) 10.2 ± 3.5 vs. intravenous glucose or (2) 0.54 ± 0.24 (2) 8.69 ± 1.52 end of each 1.7 ± 2.6 (p \ 0.001) (1) Mean 9.6 ± 3.4 glucagon injection, (2) Mean (± SD) vs. 0.53 ± 0.20 vs. 8.70 ± 1.65 treatment HbA1c (p = 0.005) number of patients (3) 4.02 ± 4.5 vs. daily dose of (p = NS) (p = NS) period (± SD) at 4.37 ± 4.5 short-and rapid- (3) 9.7 ± 4.0 vs. (2) Nocturnal baseline (1) Mean (p = 0.023) acting insulin 10.6 ± 4.3 hypoglycemia: from (± SD) BG (2) Mean (p \ 0.001) midnight to 06:00 h concentrations HbA1c at at 03:00 h (4) 8.6 ± 3.5 vs. (3) Any hypoglycemia: endpoint 9.3 ± 3.7 symptoms present or (2) Mean fasting (p = 0.003) measured BG BG \ 3.0 mmol/L, concentrations (5) 8.1 ± 3.4 vs. number of episodes per 8.5 ± 3.4 (3) Mean BG 2-h patient per month (p = NS) post-breakfast (4) Mean BG 2-h post-dinner (5) Mean BG 2-h post-mid-day meal Diabetes Ther (2018) 9:891–917 903 Table 2 continued First Analog HbA1c outcomes BG levels Hypoglycemia Secondary outcomes Author, (n) vs. Definition Results Definition Results (mmol/L) Definition Results Measures Results year comparator (n) Deeb, 2001 Lispro before Tested at (1) 8.40 ± 1.1 vs. 7-point SMBG (1) 9.6 ± 1.6 vs. Self-reported (1) 2 vs. 3 vs. 6 Mean total daily 0.21 vs. 0.23 vs. [24] meals baseline and 8.54 ± 1.0 vs. profiles on 10.1 ± 1.6 vs. insulin dose, 0.25 (p = NS) (1) Severe hypo: needing (p = NS) (analog 1, 3-monthly 8.43 ± 1.0 2 days at 10.0 ± 1.7 units/kg third-party assistance, 53) vs. during study (p = NS) baseline and 5/108 vs. 6/57 (p = 0.007 analog resulting in a coma, or lispro after end of each (1) Mean 1 vs. analog 2 requiring intravenous (2) 14.7 ± 11.9 vs. meals treatment (± SD) (p = 0.024 for glucose or glucagon, 13.6 ± 9.3 vs. (analog 2, period HbA1c at trend) number of patients 13.8 ± 9.8 (p = NS) 55) vs. endpoint (1) Overall mean RHI (57) (2) 11.7 ± 4.4, (2) Any hypoglycemia: BG (± SD) 13.5 ± 5.5, symptomatic or measured (2) Mean BG 2-h 15.0 ± 5.4 BG \ 3.5 mmol/L, post-breakfast (p \ 0.001 number of episodes per analog 1 vs. 30 days (3) Mean BG comparator; pre-lunch p = 0.023 analog 1 (4) Mean BG 2-h vs. analog 2) post-lunch (3) 8.7 ± 3.9 vs. (5) Mean BG 2-h 8.3 ± 3.1 vs. post-dinner BG 8.3 ± 3.1 vs. 9.5 ± 4.1 (p = NS analog 1 vs. analog 2, p = 0.037 analog 1 vs. RHI) (4) Analog 1 = analog 2 = RHI (p = NS) (5) 8.8 ± 5.0 vs. 9.9 ± 4.7 vs. 10.8 ± 5.4 (p = NS analog 1 vs. analog 2, p = 0.006 analog 1 vs. RHI) 904 Diabetes Ther (2018) 9:891–917 Table 2 continued First Analog HbA1c outcomes BG levels Hypoglycemia Secondary outcomes Author, (n) vs. Definition Results Definition Results (mmol/L) Definition Results Measures Results year comparator (n) Tupola, Lispro (22) Tested at (1) 0.2 ± 0.8 vs. 7-point SMBG (1) 11.5 ± 4.5 vs. Self-reported and BG meter (1) 2 vs. 2 (p = NS) (1) Total daily (1) Analog = RHI 2001 vs. RHI baseline and - 0.4 ± 0.7 profiles on 8.4 ± 3.8 analysis insulin dose (p = NS) (2) 34 vs. 41 (p = 0.6) [30] (22) 3-monthly 1 day per (p = 0.02) (p = 0.1) (1) Severe hypo: loss of (2) Patient (2) 18 (82) vs. 4 during study month over (3) 4.9 vs. 4.4 (p = 0.3) (2) 11.7 ± 6.0 vs. consciousness during a satisfaction: (18) study period (1) Mean 9.6 ± 5.7 hypoglycemic episode number (%) of (± SD) (1) Mean BG (p = 0.4) patients/families (2) Nocturnal change in (± SD) pre- who wanted to (3) 11.5 ± 5.0 vs. hypoglycemia: from 11:00 HbA1c from breakfast continue this 10.6 ± 6.0 to 06:00 h baseline to type of insulin (2) Mean BG (p = 0.8) endpoint (3) Hypoglycemia: pre-dinner (4) Analog = RHI symptoms present or (3) Mean BG at (p = NS) BG \ 3 mmol/L, bedtime number of episodes per patient per month (4) 1-h and 2-h postprandial BG excursions Danne, Aspart (23) Tested at (1) Analog = RHI 7-point SMBG (1) Analog = RHI Self-reported (1) 2 vs. 3 WHO–DTSQs (1) 4.8 ± 1.5 vs. 2007 vs. RHI baseline and (p = NS) during study (p = NS) 3.7 ± 1.8 (1) Major hypoglycemia: (2) 2.75 vs. 2.74 (1) Mean score [23] (25) 3-monthly period those that parents could (± SD) for (p = 0.045) during study (3) 1.06 (0.96;1.17) (1) Daily BG not handle on their own, question on (p = 0.225) (2) 5.0 ± 1.4 vs. (1) HbA1c at variations number of episodes continuing 4.2 ± 1.8 endpoint present form of (2) Any hypoglycemia, insulin (p = 0.051) number of episodes per week (2) Mean score (± SD) for (3) Relative risk of question on hypoglycemia (95% CI): recommending relative risk for analog/ this form of RHI insulin to others ADA American Diabetes Association, AUC area under the curve, CGM continuous glucose monitoring, CNS central nervous system, DTSQ Diabetes Treatment Satisfaction Questionnaire, DKA diabetic ketoacidosis, EM evening meal, WHO World Health Organization Diabetes Ther (2018) 9:891–917 905 906 Diabetes Ther (2018) 9:891–917 bFig. 2 Meta-analyses of key outcomes. a Forest plot there were no significant differences in FBG or showing the difference in risk of severe hypoglycemic rates of hyper- and hypoglycemia. However, the episodes with insulin analog treatment compared to regular daily insulin dose was significantly lower for human insulin (RHI) in a pediatric population. b Forest groups on insulin aspart (0.86 ± 0.237 vs. plot showing the difference in the mean fasting blood 0.94 ± 0.233 U/kg, for insulin aspart vs. insulin glucose level with insulin analog treatment compared to lispro, respectively; p = 0.018) [43]. In two RHI treatment in the CSII sub-review. c Forest plot related, 24-week, randomized, crossover trials in showing the difference in mean postprandial blood glucose adults with T1D, insulin lispro was assessed to (BG) level with insulin analog treatment compared to RHI be non-inferior to insulin aspart based on SMBG treatment in the CSII sub-review. d Forest plot showing profiles averaged over days 1–6 of treatment, the difference in risk of severe hypoglycemic episodes with but not when day 6 values alone were compared insulin analog treatment compared to RHI treatment in [39]. In a randomized, parallel-group trial in 146 the CSII sub-review. e Forest plot showing the mean adults with T1D, mean change from baseline difference in the rate of any hypoglycemic episodes with was not significantly different for participants insulin analog treatment compared to RHI treatment in treated with insulin lispro, insulin aspart or RHI the CSII sub-review. f Forest plot showing the difference for 16 weeks [31]. Rates of hypoglycemia were in glycated hemogloblin (HbA1c) with insulin analog also similar among treatments. treatment (lispro or aspart) compared to RHI in the CSII One RCT compared all three RAIAs in adults sub-review. [40]. This was a crossover trial with three Squares and diamonds represent the difference in HbA1c 13-week periods that was designed to test the after intervention with the two treatments for each study superiority of insulin glulisine for unexplained (horizontal lines are 95% CI) and for all the studies combined, respectively. The I value refers to the statistical hyperglycemia and/or infusion-set occlusion. It heterogeneity for the pooled analysis. A random-effects failed to show superiority of insulin glulisine on model using generic inverse variance showed a mean the primary outcome, but revealed that the difference in HbA1c of - 0.19% (95% CI - 0.46; 0.08); monthly rate of unexplained hyperglycemic - 2.1 (95% CI - 5.0; 0.9) mmol/mol with insulin analog episodes and/or perceived catheter-set occlu- compared to RHI at the end of the treatment period. The sion was significantly higher in insulin gluli- squares and the diamond in a, d, e represent the difference sine-treated patients than in those receiving the in risk for each study (horizontal lines represent 95% CI) two other analogs [40]. Furthermore, insulin and for all studies combined, respectively. The squares and glulisine was associated with a higher frequency the diamond in b, c represent the difference in the glucose of symptomatic hypoglycemia, whereas HbA1c levels between the two treatment arms for each study and 7-point SMBG were similar for all three (horizontal lines are 95% CI) and for all the studies insulin analogs [40]. combined, respectively. The results of these meta-analyses are the mean of post-breakfast BG measurements only. In Meta-analysis a and d ‘Events’ refers to the number of patients Data on mean FBG and mean PPG for patients experiencing any such episode during the treatment period using CSII were sufficiently consistent to permit as a proportion of total number of patients in that some meta-analysis, as were some of the hypo- treatment group. ‘Rate’ refers to mean (± SD) of any glycemic outcomes. Funnel plots suggested that episodes of hypoglycemia per 30 days in all the patients in there was no publication bias, although the the respective treatment group. In f ‘Bode, 2002 (a)’ [31] refers to the observed difference in HbA1c between the number of studies in the meta-analysis with subgroup of insulin lispro vs. RHI, and ‘Bode, 2002 (b)’ FBG outcomes was small (funnel plots not [31] refers to the subgroup on insulin aspart vs. RHI; the shown). Meta-analysis was performed for RAIAs three remaining studies compare lispro vs. RHI. The ‘I ’ versus RHI for FBG (three studies) (Fig. 2b), PPG value refers to the statistical heterogeneity for this pooled (five studies) (Fig. 2c), severe hypoglycemic analysis. CI confidence interval, IV inverse variance, episodes (six studies) (Fig. 2d), any hypo- SD standard deviation glycemic episodes (five studies) (Fig. 2e) and HbA1c (four studies) (Fig. 2f). A random-effects model using generic inverse variance showed a mean difference in FBG of - 0.53 mmol/L Diabetes Ther (2018) 9:891–917 907 Table 3 Study outcomes: continuous subcutaneous insulin infusion First author, Analog HbA1c outcomes BG levels Hypoglycemia Secondary outcomes year (n) vs. Definition Results Definition Results (mmol/L) Definition Results Measures Results comparator (n) Zinmann, Lispro (30) Tested monthly (1) 8.03 ± 0.13 8 point-SMBG (1) No significant Self-reported diary (1) 12.7 ± 1.6 (1) 5-h meal test on (1) Peak free plasma insulin at 1997 [41] vs. RHI profile done difference between subgroup of six 45 min for lispro (1) Mean (2) 7.66 ± 0.13 vs. (1) Any hypo at (2) 8.6 ± 1.4 vs. (30) weekly treatments patients for (287 ± 69 pmol/L) vs. (± SE) 8.00 ± 0.16 baseline: 10.8 ± 1.8 postprandial 150 min for RHI (294 ± 56). baseline (1) Mean (2) Lower with lispro than BG \ 3 mmol/L or (lispro (3) - 0.34 insulin glucose Plasma glucose and FFA HbA1c in all fasting BG with RHI (p \ 0.05) compatible p = 0.035 vs. (p = 0.0041) and FFA markedly reduced for lispro vs. patients symptoms, episodes baseline, RHI (2) Mean post- RHI per 30 days (± SE) p = NS) (2) BW (2) Mean prandial BG (2) No difference in BW from endpoint (2) Any hypo during (3) 8.4 ± 1.3 (3) Infusion-set baseline or between treatments HbA1c treatment occlusion (4) 6.0 ± 0.9 vs. at endpoint (3) Difference in (3) Biochemical hypo 7.6 ± 1.3 (3) No occlusions with either endpoint at baseline: (lispro treatment HbA1c BG \ 3 mmol/L p = 0.03 vs. between only baseline, RHI treatments p = NS) (4) Biochemical hypo during treatment (5) 0/30 vs. 0/30 (5) Severe hypo Melki, 1998 Lispro (38) Tested at end of (1) 7.74 ± 0.20 vs. 7 point-SMBG (1) 7.93 ± 0.15 vs. Self-reported diary (1) 7.03 ± 0.94 (1) Ketoacidosis (1) No episodes in either group [35] vs. RHI first 7.97 ± 0.13 profile done 8.61 ± 0.18 recorded in last vs. (2) BW (2) No difference (38) 3 months daily during (p \ 0.0001) 30 days of first 7.94 ± 0.88 (2) 7.11 ± 0.15 vs. last 30 days treatment period (p = NS) (3) Glucose (3) Significantly lower overall (1) Baseline 7.88 ± 0.16 (2) 7.70 ± 0.17 vs. of each variability mean and postprandial glycemic 7.75 ± 0.21 (p = NS) (1) Hypo event: (2) 0.05 ± 0.05 (2) Endpoint (3) - 0.62 ± 0.13 treatment fluctuation with lispro than BG \ 3 mmol/L, vs. 0.47 ± 0.19 (4) Patient vs. - 0.09 ± period (3) 8.26 ± 0.19 vs. RHI, SD of BG in mmol/L (3) Change from episodes per month (p \ 0.05) preference 0.15 (p = 0.01) 9.90 ± 0.20 (± SE): 3.44 ± 0.10 vs. baseline to (1) Mean BG (± SD) (p \ 0.0001) (3) 3 vs. 7 (3/38 (5) Technical 3.80 ± 0.10 (p \ 0.001) and endpoint ± SE (2) Very low BG vs. 4/38) problems 3.58 ± 0.10 mmol/L vs. (2) Mean measurement: 3.84 ± 0.10 (p \ 0.02) preprandial BG \ 2 mmol/L (4) All seven questions on non- BG (3) Severe hypo: third- validated questionnaire in favor (3) Mean 2-h party assistance of lispro (p \ 0.0001) postprandial required (5) Insulin precipitation in BG catheter, one vs. four episodes; catheter obstruction, nine episodes each 908 Diabetes Ther (2018) 9:891–917 Table 3 continued First author, Analog HbA1c outcomes BG levels Hypoglycemia Secondary outcomes year (n) vs. Definition Results Definition Results (mmol/L) Definition Results Measures Results comparator (n) Schmauss, Lispro (11) Tested Lispro then RHI vs. 3-day BG (1) 6.5 ± 0.4 vs. Self-reported (1) 4.0 ± 0.9 vs. (1) Basal and bolus (1) No difference 1998 [38] vs. RHI 3-monthly RHI then lispro profile done 7.5 ± 0.4 3.2 ± 0.7 insulin (1) Hypo: (2) No change in BMI (11) 3-monthly (p = NS) requirements (1) Baseline (1) 6.3 ± 0.2 vs. (p = NS) BG \ 3.5 mmol/L (3) No difference in treatment 6.7 ± 0.4 (1) Mean and/or symptoms, (2) 0 vs. 0 (0/11 (2) BMI change (2) End of first (2) 6.8 ± 0.3 vs. satisfaction fasting BG episodes per 30 days vs. 0/11) 3 months (p = NS) 8.3 ± 0.3 (3) Treatment (± SD) (2) Mean 2-h satisfaction (3) End of study (2) 5.7 ± 0.3 vs. (p = 0.03) postprandial (2) Severe hypo: 6.5 ± 0.3 BG requiring IV glucose (p = NS) or glucagon (3) 6.2 ± 0.2 vs. 6.3 ± 0.3 (p = NS) Guerci, 1999 Lispro (10) Tested monthly (1) 7.17 ± 0.86 vs. SMBG done (1) 9.35 ± 1.17 vs. Self-reported (1) 10.1 ± 9.7 vs. (1) Glucose (1) No difference between [32] vs. RHI 7.36 ± 0.76 monthly 9.07 ± 0.43 (p = NS) 6.9 ± 4.4 variability, mean treatment groups (1) Baseline (1) Incidence of hypo (10) (p = NS) (p = NS) SD of previous (1) Mean (2) 9.04 ± 0.89 vs. at baseline: (2) Until 3 h no difference in PG (2) Endpoint of months’ BG (2) 7.07 ± 0.51 vs. baseline BG 9.32 ± 1.17 (p = NS) BG \ 3.5 mmol/L, (2) 7.1 ± 4.6 vs. between groups. From 3 to 5 h, each treatment 6.97 ± 0.67 episodes per 30 days 12.6 ± 10.2 After pump PG higher with lispro (p \ 0.01) period (2) Mean BG (3) 9.53 ± 1.98 vs. (p = NS) (± SD) (p = 0.05) interruption for at endpoint 9.92 ± 1.05 (p = NS) (3) From 3 h onwards, consistently 5h: (2) Incidence of hypo higher for lispro vs. RHI (but (3) Mean (4) 9.43 ± 1.39 vs. at endpoint (2) Mean PG p = NS) baseline 10.49 ± 2.05 postprandial (p = 0.05) (3) Plasma (4) From 2 h onwards, consistently BG 3-hydroxybuturate higher for lispro vs. RHI (p \ 0.05) (4) Mean (4) Plasma FFA postprandial BG at endpoint Diabetes Ther (2018) 9:891–917 909 Table 3 continued First author, Analog HbA1c outcomes BG levels Hypoglycemia Secondary outcomes year (n) vs. Definition Results Definition Results (mmol/L) Definition Results Measures Results comparator (n) Renner, 1999 Lispro (113) Tested at (1) 7.24 ± 1.0 8-point daily (1) 7.2 ± 1.7 vs. Self-reported (1) 12.0 ± 13.9 (1) Ketosis (1) Five vs. four patients [37] vs. RHI baseline and SMBG 7.8 ± 2.1 vs. (2) 6.77 ± 0.88 vs. (1) BG \ 3.5 mmol/L (2) Treatment (2) Patients scored significantly (113) endpoint 11.0 ± 11.2 6.90 ± 0.97 (1) Mean (p = NS) and/or symptoms, satisfaction higher on the treatment (p = NS) (1) Baseline in (p = 0.02) fasting BG mean number of (DTSQ) satisfaction when treated with (2) 7.0 ± 1.9 vs. all patients (± SD) episodes (± SD) per (2) Lispro = RHI lispro compared to RHI 8.6 ± 2.6 (p \ 0.001) (3) Occlusion of patient per 4 months (p = NS) (2) Endpoint (2) Mean post- catheter (3) Same amount in the two (3) 6.9 ± 1.9 vs. breakfast BG (2) BG \ 3.3 mmol/L, treatments 7.3 ± 2.2 mean number of (3) Mean pre- (p = NS) episodes (± SD) per lunch BG patient per month (4) 7.6 ± 1.9 vs. (4) Mean post- 8.7 ± 2.4 lunch BG (p \ 0.001) (5) Mean pre- dinner BG (5) 7.3 ± 1.9 vs. 7.5 ± 1.9 (6) Mean post- dinner BG (p = NS) (7) Mean (p \ 0.001) 22:00 h BG (6) 7.2 ± 1.9 vs. (8) Mean 8.3 ± 1.9 02:00 h BG (7) 7.6 ± 1.8 vs. 8.3 ± 2.0 (p \ 0.001) (8) 8.0 ± 2.7 vs. 7.7 ± 2.3 (p = NS) Johansson, Lispro (41) Tested at (1) 7.7 ± 0.8 Last 30 days (1) 8.5 vs. 8.4 (p = NS) Self-reported (1) 9.7 vs. 8.0 [1.7 (1) Treatment (1) No difference between the 2000 [34] vs. RHI baseline and SMBG (- 1.3; 5.3)] satisfaction treatments (2) 7.4 vs. 7.6 (2) 8.1 vs. 9.6 (p \ 0.001) (1) Any hypo: (41) endpoint (p = NS) (DTSQ) (1) Mean of BG \ 3.0 mmol/L (p = 0.047) (3) 8.3 vs. 8.9 (p \ 0.001) (1) Baseline in pre-prandial and/or symptoms, all patients (3) - 0.2 (- 0.3; and bedtime episodes per 30 days 0.0) BG (difference between (2) Endpoint treatments [95% (p = 0.047) (2) Mean of (3) Difference CI]) postprandial between BG treatments (95% CI) (3) Mean of all SMBG 910 Diabetes Ther (2018) 9:891–917 Table 3 continued First author, Analog HbA1c outcomes BG levels Hypoglycemia Secondary outcomes year (n) vs. Definition Results Definition Results (mmol/L) Definition Results Measures Results comparator (n) Raskin, 2001 Lispro (58) Tested at (1) 7.9 ± 1.1 vs. 4-point daily (1) 8.1 ± 2.0 vs. Self-reported (1) 8 vs. 11 (1) BW (1) No difference between [36] vs. RHI baseline and 7.6 ± 0.8 SMBG 8.1 ± 1.6, (p = NS) (p = NS) treatments (1) Any hypo: (58) endpoint (p = NS) done in last (2) 11.16 ± 4.29 vs. BG \ 3.0 mmol/L (2) 3 vs. 3 2 weeks of (1) Baseline in (2) - 0.34 ± 13.20 ± 4.68 and/or symptoms, (p = NS) each all patients 0.59 vs. (p = 0.012) episodes per treatment 3/58 vs. 2/58 - 0.09 ± 12 weeks (2) Change from period (3) 9.64 ± 4.10 vs. 0.63 (p = 0.004) baseline 12.53 ± 4.64 (2) Severe (1) Mean BG (p = 0.001) hypoglycemia: (± SD) requiring IV glucose PG during test meal at end of each treatment period (2) Mean PG 1 h post- meal (3) Mean PG 2 h post- meal Tubiani-Rufi, Lispro (27) Tested at (1) ? 0.15 ± 0.13 9-point SMBG (1) - 0.94 ± 5.05 vs. ? Self-reported (1) 2 vs. 2 (1) BW (1) No difference between 2004 [42] vs. RHI baseline and vs. ? on 2 days 1.78 ± 5.94 (p = 0.01) treatments (1) Severe 2/27 vs. 2/27 (2) Glucose (27) endpoint 0.11 ± 0.63 during each hypoglycemia, variability (SD of (2) 6.4 ± 1.0 vs. (p = NS) treatment (2) 11.0 ± 6.4 vs. (1) Difference in episodes per 30 days SMBG) 6.0 ± 0.95 mmol/L (p = 0.01) 13.8 ± 8.5 HbA1c (1) Change in (2) BG B 3.3 mmol/L (p = NS) (3) Infusion-set (3) No difference between (± SD) from mean post- occlusion treatments baseline to end dinner BG (3) BG B 2.2 mmol/L (3) 0.6 ± 1.1 vs. of first 1.0 ± 1.1 (4) Parents’ (4) 74% want to continue lispro treatment (p = NS) treatment (5) 2.5 ± 2.7 vs. 2.0 ± 3.0 sequence preference episodes (p = NS) (5) Hyperglycemic episodes with ketonuria Diabetes Ther (2018) 9:891–917 911 Table 3 continued First author, Analog HbA1c outcomes BG levels Hypoglycemia Secondary outcomes year (n) vs. Definition Results Definition Results (mmol/L) Definition Results Measures Results comparator (n) Hoogma, Glulisine Tested at (1) 6.8 vs. 7.1 NR NR Self-reported (1) 2 vs. 2 (1) Infusion-set (1) No difference between 2006 [33] (29) vs. baseline and (p = NS) occlusion treatments (1) Severe (2) 20 vs. 15 aspart endpoint (2) 7.0 vs. 7.2 hypoglycemia: (p = NS) (2) Unexplained (2) 6 vs. 12 (p = NS) (30) (1) Baseline BG \ 2.0 mmol/L hyperglycemia: at (3) 0.11 (3) 26 vs. 24 HbA1c and requiring least one episodes (p = NS) (- 0.09; 0.31) assistance or IV of (2) Endpoint (p = NS) glucose/glucagon, BG [ 19.4 mmol/ HbA1c number of patients L, episodes in (3) Between- experiencing at least 12 weeks group one in 12 weeks difference in (2) Nocturnal HbA1c hypoglycemia (95% CI) from baseline (3) Symptomatic to endpoint hypoglycemia Weinzimer, Aspart (198) Tested at (1) 8.0 ± 0.94 vs. 8-point SMBG (1) 9.5 ± 4.3 vs. Self-reported (1) 0.4 vs. 0.3 (1) BW (1) No difference between 2008 [43] vs. lispro baseline and 8.2 ± 0.84 done on 9.9 ± 3.8 (p = NS) (p = NS) treatments (1) Major (2) Hyperglycemia: (100) endpoint (p = NS) 2 days at (2) 9.3 ± 3.7 vs. hypoglycemia: (2) 5.7 vs. 6.2 BG [ 16.7 mmol/ (2) 11 vs. 17 (p = NS) baseline and (1) Baseline (2) - 0.15 ± 10.0 ± 4.6 BG \ 3.1 mmol/L (p = NS) L, % patients with before (3) 1 vs. 2 (p = NS) 0.05 vs. and requiring episodes (2) Change in ending (p = NS) (3) 77.2 vs. 66.0 - 0.05 ± 0.07 assistance or IV HbA1c from treatment (p = NS) (3) DKA, number of (p = NS) (3) aspart = lispro glucose/glucagon, baseline to end patients with (1) Mean (p = NS) episodes per patient of treatment (3) 50.3 vs. 40.4% episodes fasting BG per year (p = NS) (4) aspart = lispro (3) Percentage (± SD) at (p = NS) (2) Nocturnal achieving age- (4) 59.7 vs. 43.8% baseline hypoglycemia: minor specific (p = 0.04) (2) Mean or major occurring HbA1c target fasting BG at between midnight at baseline endpoint and 06:00 h (4) Percentage (3) Mean (3) Minor achieving age- postprandial hypoglycemia: specific BG at BG \ 3.1 mmol/L HbA1c target endpoint with or without at endpoint symptoms (4) Mean BG at endpoint 912 Diabetes Ther (2018) 9:891–917 Table 3 continued First author, Analog HbA1c outcomes BG levels Hypoglycemia Secondary outcomes year (n) vs. Definition Results Definition Results (mmol/L) Definition Results Measures Results comparator (n) Tamborlane, Substudy 1 Tested at Substudy 1 7-point SMBG Substudy 1 Self-reported (1) 5 vs. 16 (1) Infusion-set (1) No difference between 2015 [39] baseline and done on occlusion treatments lispro (116) (1) 7.3 ± 0.7 (1) 9.34 ± 0.14 vs. (1) Severe hypo: (2) lispro = endpoint day 6 of vs. aspart 9.11 ± 0.14 (p [ 0.05) requiring third-party aspart (2) Hyperglycemia: (2) Substudy 1: 8.20 vs. 6.79 (2) 7.34 ± 0.05 vs. reservoir use (116) (1) Mean assistance, total BG[ 13.9 (p = 0.029) 7.28 ± 0.05 Substudy 2 (3) Substudy 1: (± SD) (1) Mean number of episodes mmol/L ([3h Substudy 2 (p = NS) 9.39 vs. 10.84 Substudy 2: 8.05 vs. 6.54 HbA1c at (± SEM) (1) 8.72 ± 0.10 vs. after eating) (2) Nocturnal hypo: (p = 0.003) (p = 0.003) lispro (118) baseline (3) 0.06 ± 0.05 vs. BG 8.54 ± 0.10 (p [ 0.05) or[ 16.7 mmol/L occurring between vs. aspart - 0.00 ± 0.05 Substudy 2: 7.57 (within 3 h after (3) 4 vs. 0 (2) Mean bedtime and wake up (118) (p = NS) vs. 8.71 eating), episodes per (± SEM) (3) Symptomatic hypo: (p = 0.012) 30 days endpoint Substudy 2 BG \ 3.9 mmol/L HbA1c (4) Substudy 1: (3) DKA, number of (1) 7.5 ± 0.7 and symptomatic, 15.26 vs. 16.91 episodes (3) Mean change episodes per 30 days (2) 7.30 ± 0.04 vs. (p = 0.006) (± SEM) in 7.14 ± 0.04 (4) Symptomatic hypo: HbA1c from Substudy 2: 16.74 (p \ 0.001) total number of baseline vs. 18.86 episodes (3) - 0.15 vs. (p \ 0.001) - 0.31 (5) Asymptomatic: (5) BG \ 3.9 mmol/L (p \ 0.001) lispro = aspart and no symptoms Bode, 2002 Aspart (59) Tested at (1) 7.3 ± 0.7 vs. 8-point SMBG (1) Aspart = lispro = Self-reported (1) 0 vs. 0 vs. 1 (1) BW (1) No difference in any treatment [31] vs. lispro baseline and 7.3 ± 0.7 vs. done on RHI group (1) Severe hypo: 0/59 vs. 1/59; (2) Unexplained (28) vs. endpoint 7.5 ± 0.8 2 days at (2) BG \ 2.8 mmol/L 0/28 vs. 1/59 hyperglycemia: (2) 27% vs. 36% vs. 41%, RHI (59) (p = NS) baseline and (1) Mean Aspart \ lispro = RHI and requiring BG [ 19.4 mmol/ endpoint (2) 0.5 ± 0.83 vs. (p = NS) (± SD) (2) 0.00 ± 0.51 vs. assistance or IV L, % patients with (p = 0.019) 0.6 ± 0.61 vs. HbA1c at 0.18 ± 0.84 vs. (1) Fasting and glucose/glucagon, at least one episode (3) No episodes 0.9 ± 0.97 baseline 0.15 ± 0.63 pre-meal BG (3) RHI \ aspart = lispro mean events per (aspart vs. (3) DKA (p = NS) patient per 30 days (2) Mean change (2) Post-dinner (p = 0.002) lispro p = NS, (± SD) from baseline BG aspart vs. RHI to endpoint (2) Nocturnal hypo: p = 0.004) (3) BG at minor or major 02:00 h (3) 3.7 ± 3.6 vs. occurring between 4.4 ± 4.7 vs. midnight and 4.8 ± 4.2 06:00 h (p = NS) (3) Minor hypo: (4) 6.7 ± 5.4 vs. symptomatic and 10.5 ± 8.1 vs. BG \ 2.8 mmol/L 10.5 ± 8.9 (4) Minor hypo: (aspart vs. symptoms with or lispro without low BG p = 0.044, aspart vs. RHI p = 0.034) Diabetes Ther (2018) 9:891–917 913 Table 3 continued First author, Analog HbA1c outcomes BG levels Hypoglycemia Secondary outcomes year (n) vs. Definition Results Definition Results (mmol/L) Definition Results Measures Results comparator (n) van Bon, Glulisine Tested at (1) 7.31 ± 0.71 vs. 7-point SMBG (1) glulisine = Self-reported (1) 1.63 ± 4.50 (1) Infusion-set (1) No difference between groups 2011 [40] (256) vs. baseline and 7.33 ± 0.71 vs. 1 day before aspart = RHI vs. occlusion (1) Severe (2) 1 vs. 0 vs. 0 aspart endpoint 7.28 ± 0.71 endpoint 1.38 ± 7.73 (2) (glulisine vs. aspart) 9.2 hypoglycemia: (2) DKA (256) vs. (p = NS) vs. (3) 0.14 ± 0.43 vs. 0.06 ± 0.22 (1) Mean (1) Fasting and vs. 8.6 (p \ 0.021) BG \ 2.8 mmol/L lispro 1.06 ± 3.81 (3) Hyperketonemia, vs. 0.06 ± 0.18 (glulisine vs. (± SD) (2) 7.32 ± 0.73 vs. pre-meal and requiring (256) (3) (glulisine vs. lispro) 8.8 rate per month aspart HbA1c at 7.26 ± 0.76 vs. SMBG assistance or IV (p = NS) vs. 8.2 (p \ 0.018) (± SD) baseline 7.31 ± 0.74 glucose/glucagon, p = 0.01, glulisine vs. lispro (2) Mean post- (2) 0.61 ± 2.20 (p = NS) episodes per patient- p = 0.02) (2) Mean lunch BG vs. 0.33 ± 1.18 year (± SD) HbA1c at (3) 28 vs. 31 vs. 30 vs. 0.36 ± 1.40 (3) Mean endpoint (p = NS) (2) Nocturnal severe (glulisine vs. nocturnal hypo: severe hypo aspart (3) Percentage of BG occurring at night patients with p = 0.044, HbA1c \ 7.0 (3) Symptomatic hypo: glulisine vs. at endpoint, % BG \ 3.9 mmol/L lispro and symptomatic p = 0.070) (4) Nocturnal (3) 73.84 ± symptomatic hypo: 82.10 vs. symptomatic hypo 65.01 ± 72.07 occurring at night vs. 62.69 ± 72.87 (glulisine vs. aspart p = 0.008, glulisine vs. lispro p \ 0.001) (4) 12.80 ± 18.02 vs. 9.66 ± 13.75 vs. 9.48 ± 13.28 (glulisine vs. aspart p \ 0.001, glulisine vs. lispro p \ 0.001) BMI Body mass index, CI confidence interval, CSII continuous subcutaneous insulin infusion, DKA diabetic ketoacidosis, FFA free fatty acids, IV intravenous, PG plasma glucose, SE standard error, Aspart insulin aspart, BG blood glucose, BMI body mass index, BW body weight, CI confidence interval, CSII continuous subcutaneous insulin infusion, DKA diabetic ketoacidosis, DTSQ Diabetes Treatment Satisfaction Questionnaire, FFA free fatty acids, glulisine insulin glulisine, IV intravenous, hypo hypoglycemia, lispro insulin lispro, NR not reported, NS not statistically significant, PG plasma glucose, RHI regular human insulin, SMBG self-measured blood glucose 914 Diabetes Ther (2018) 9:891–917 (95% CI - 1.21; 0.15), a mean difference for analysis was possible, primarily in the CSII PPG of – 1.63 mmol/L (95% CI - 1.71; –1.54), a population. Those results indicate that RAIAs in risk difference for severe hypoglycemic episodes CSII lower post-breakfast BG and possibly of - 0.01 (95% CI - 0.04; 0.02) and a mean HbA1c to a greater extent than RHI, without an difference in rate of any hypoglycemic episode increased risk of hypoglycemia. Although many of - 0.75 (95% CI - 2.21; 0.72). The mean dif- trials have been published using CSII versus ference in HbA1c was - 0.19% (95% CI - 0.46; multiple daily injections (MDI) in pediatric 0.08); - 2.1 mmol/mol (95% CI - 5.0; 0.9) with subjects, we did not review those here because it RAIAs compared to RHI after 3 or 4 months of was impossible to separate the effects of the treatment. RAIA from those of the mode of treatment (CSII or MDI). DISCUSSION CONCLUSIONS This systematic review and meta-analysis sum- marizes the safety and efficacy of RAIAs in Rapid-acting insulin analogs appear to be safe populations of patients who are either typically and effective in these three special populations excluded from clinical trials (i.e. due to preg- of people with T1D. However, additional trials nancy) or require dedicated trials (i.e. children would be helpful, and head-to-head trials would and adolescents, patients using CSII). Overall, be necessary to detect any statistical differences for insulin lispro and insulin aspart, data across among them, should they exist. The lack of all three special populations indicate that their clinically relevant differences in performance safety and efficacy are comparable with, and in among the RAIAs make other factors, such as some cases significantly better than, RHI. Data cost, availability and patient/provider prefer- also suggest, from the more limited results ence, more important. Finally, studies have now available, similar characteristics for insulin been published indicating the PK advantages of glulisine. a new formulation of insulin aspart in clinical There are no head-to-head RCTs individually development (faster aspart) over conventional comparing all three RAIAs with each other in insulin aspart [46], including in pediatric pop- the pediatric CSII setting. However, in 2009, the ulations [47] and in those using CSII [48]. Institute for Quality and Efficiency in Health Studies addressing the performance of new fast- Care in Germany, acknowledging the limited acting insulin aspart versus insulin glulisine and amount of data at the time, concluded that insulin lispro in these special populations in a there were no significant differences between clinical setting would be advantageous. RHI and any of the three RAIAs in terms of key efficacy and safety endpoints [44]. The UK National Institute for Health and Care Excel- ACKNOWLEDGEMENTS lence (NICE) guidelines specifically indicate that RAIAs are preferred over RHI for use in CSII for pediatric patients [45]. Funding. The authors received no funding Given the lack of any major safety concerns for data abstraction, meta-analysis and writing. when the three RAIAs were studied individually, This work was supported by Novo Nordisk A/S. there is no a priori reason to suspect that head- Novo Nordisk A/S funded the writing and edit- to-head trials would reveal any substantive dif- ing assistance as well as the article publication ferences in safety among them in special pop- charge, and reviewed the manuscript for scien- ulations. However, without randomized tific accuracy. The authors determined which comparative trials in these special populations, studies were eligible for the review and meta- that conclusion remains speculative. Due to the analysis, performed the meta-analysis and made limited number of studies and the heterogene- the decision to submit the manuscript for ity of the outcome measures, only limited meta- publication. Diabetes Ther (2018) 9:891–917 915 Medical Writing and Editorial Assis- provide a link to the Creative Commons license, tance. The authors are grateful to Gary and indicate if changes were made. Patronek, Helen Marshall and Erin Slobodian, of Watermeadow Medical, an Ashfield company, part of UDG Healthcare plc, for writing and REFERENCES editing assistance in the development of this manuscript. Gary Patronek also performed the literature search and assisted with data abstrac- 1. American Diabetes Association. Standards of medi- cal care in diabetes—2017. Diabetes Care. tion. This assistance was funded by Novo Nor- 2017;40[Suppl 1]:S1–135. disk A/S, who also had a role in the review of the manuscript for scientific accuracy. 2. Eli Lilly. Humalog prescribing information. www. accessdata.fda.gov/drugsatfda_docs/label/2013/020 563s115lbl.pdf; 2013. Accessed 29 Jan 2018. Authorship. All named authors meet the International Committee of Medical Journal 3. Novo Nordisk. NovoLog prescribing information. Editors (ICMJE) criteria for authorship for this www.accessdata.fda.gov/drugsatfda_docs/label/2015/ article, take responsibility for the integrity of 020986s082lbl.pdf; 2015. Accessed 29 Jan 2018. the work as a whole, and have given their 4. Sanofi Aventis. Apidra prescribing information. approval for this version to be published. www.accessdata.fda.gov/drugsatfda_docs/label/2015/ 021629s030lbl.pdf; 2015. Accessed 29 Jan 2018. Disclosures. Kirsten Nørgaard serves as adviser to Medtronic, Abbott and Novo Nordisk, 5. Eli Lilly. Humalog product information. www.ema. europa.eu/ema/index.jsp?curl=pages/medicines/hum owns shares in Novo Nordisk, has received an/medicines/000088/human_med_000820.jsp& research grants from Roche and Novo Nordisk mid=WC0b01ac058001d124; 2017. Accessed 29 and has received fees for speaking from Med- Jan 2018. tronic, Roche, Rubin Medical, Sanofi, Novo 6. Novo Nordisk. NovoRapid product information. Nordisk and Bayer. Ponnusamy Saravanan has www.ema.europa.eu/ema/index.jsp?curl=pages/ received honoraria for serving on advisory medicines/human/medicines/000258/human_med_ boards and speaker fees for all three rapid-acting 000935.jsp&mid=WC0b01ac058001d124; 2017. insulin analog-producing companies (Novo Accessed 29 Jan 2018. Nordisk, Sanofi and Lilly). Nithya Sukumar and 7. Sanofi Aventis. Apidra product information. www. Snorri B. Rafnsson declare they have no conflict ema.europa.eu/ema/index.jsp?curl=pages/medicines/ of interest. human/medicines/000557/human_med_000648.jsp& mid=WC0b01ac058001d124; 2017. Accessed 29 Jan Compliance with Ethics Guidelines. This article is based on previously conducted studies, 8. Bohn B, Karges B, Vogel C, et al. 20 years of pedi- and does not involve any new studies of human atric benchmarking in Germany and Austria: age- or animal subjects performed by any of the dependent analysis of longitudinal follow-up in 63,967 children and adolescents with type 1 dia- authors. betes. PLoS One. 2016;11:e0160971. Data Availability. Data used in the meta- 9. Mooradian AD. Special considerations with insulin analysis are included in Fig. 1. therapy in older adults with diabetes mellitus. Drugs Aging. 2011;28:429–38. Open Access. This article is distributed 10. Kerr D, Wizemann E, Senstius J, Zacho M, Ampudia- under the terms of the Creative Commons Blasco FJ. Stability and performance of rapid-acting Attribution-NonCommercial 4.0 International insulin analogs used for continuous subcutaneous License (http://creativecommons.org/licenses/ insulin infusion: a systematic review. J Diabetes Sci by-nc/4.0/), which permits any non- Technol. 2013;7:1595–606. commercial use, distribution, and reproduction 11. Home PD. The pharmacokinetics and pharmaco- in any medium, provided you give appropriate dynamics of rapid-acting insulin analogues and credit to the original author(s) and the source, 916 Diabetes Ther (2018) 9:891–917 their clinical consequences. Diabetes Obes Metab. young children with type 1 diabetes. Diabetes Care. 2012;14:780–8. 2006;29:2311–2. 12. Chiang JL, Kirkman MS, Laffel LM, Peters AL, Type 23. Danne T, Ra ˚ stam J, Odendahl R, et al. Parental 1 Diabetes Sourcebook Authors. Type 1 diabetes preference of prandial insulin aspart compared with through the life span: a position statement of the preprandial human insulin in a basal-bolus American Diabetes Association. Diabetes Care. scheme with NPH insulin in a 12-wk crossover 2014;37:2034–54. study of preschool children with type 1 diabetes. Pediatr Diabetes. 2007;8:278–85. 13. Hod M, Damm P, Kaaja R, et al. Fetal and perinatal outcomes in type 1 diabetes pregnancy: a random- 24. Deeb LC, Holcombe JH, Brunelle R, et al. Insulin ized study comparing insulin aspart with human lispro lowers postprandial glucose in prepubertal insulin in 322 subjects. Am J Obstet Gynecol. children with diabetes. Pediatrics. 2008;198(186):e1–7. 2001;108:1175–9. 14. Mathiesen ER, Kinsley B, Amiel SA, et al. Maternal 25. Fairchild JM, Ambler GR, Genoud-Lawton CH, et al. glycemic control and hypoglycemia in type 1 dia- Insulin lispro versus regular insulin in children with betic pregnancy: a randomized trial of insulin type 1 diabetes on twice daily insulin. Pediatr Dia- aspart versus human insulin in 322 pregnant betes. 2000;1:135–41. women. Diabetes Care. 2007;30:771–6. 26. Ford-Adams ME, Murphy NP, Moore EJ, et al. 15. Persson B, Swahn ML, Hjertberg R, et al. Insulin Insulin lispro: a potential role in preventing noc- lispro therapy in pregnancies complicated by type 1 turnal hypoglycaemia in young children with dia- diabetes mellitus. Diabetes Res Clin Pract. betes mellitus. Diabet Med. 2003;20:656–60. 2002;58:115–21. 27. Holcombe JH, Zalani S, Arora VK, Mast CJ, Lispro in 16. Di Cianni G, Volpe L, Ghio A, et al. Maternal Adolescents Study Group. Comparison of insulin metabolic control and perinatal outcome in women lispro with regular human insulin for the treatment with gestational diabetes mellitus treated with lis- of type 1 diabetes in adolescents. Clin Ther. pro or aspart insulin: comparison with regular 2002;24:629–38. insulin. Diabetes Care. 2007;30:e11. 28. Pan ´ kowska E, Nazim J, Szalecki M, Urban M. Equal 17. Mecacci F, Carignani L, Cioni R, et al. Maternal metabolic control but superior caregiver treatment metabolic control and perinatal outcome in women satisfaction with insulin aspart in preschool chil- with gestational diabetes treated with regular or dren. Diabetes Technol Ther. 2010;12:413–8. lispro insulin: comparison with non-diabetic preg- nant women. Eur J Obstet Gynecol Reprod Biol. 29. Philotheou A, Arslanian S, Blatniczky L, Peterkova 2003;111:19–24. V, Souhami E, Danne T. Comparable efficacy and safety of insulin glulisine and insulin lispro when 18. Pettitt DJ, Ospina P, Howard C, Zisser H, Jovanovic given as part of a basal-bolus insulin regimen in a L. Efficacy, safety and lack of immunogenicity of 26-week trial in pediatric patients with type 1 dia- insulin aspart compared with regular human insu- betes. Diabetes Technol Ther. 2011;13:327–34. lin for women with gestational diabetes mellitus. Diabet Med. 2007;24:1129–35. 30. Tupola S, Komulainen J, Ja ¨a ¨skela ¨inen J, Sipila ¨ I. Post-prandial insulin lispro vs. human regular 19. Jovanovic L, Ilic S, Pettitt DJ, et al. Metabolic and insulin in prepubertal children with type 1 diabetes immunologic effects of insulin lispro in gestational mellitus. Diabet Med. 2001;18:654–8. diabetes. Diabetes Care. 1999;22:1422–7. 31. Bode B, Weinstein R, Bell D, et al. Comparison of 20. McCance DR, Damm P, Mathiesen ER, et al. Eval- insulin aspart with buffered regular insulin and uation of insulin antibodies and placental transfer insulin lispro in continuous subcutaneous insulin of insulin aspart in pregnant women with type 1 infusion: a randomized study in type 1 diabetes. diabetes mellitus. Diabetologia. 2008;51:2141–3. Diabetes Care. 2002;25:439–44. 21. Lloyd A, Townsend C, Munro V, Twena N, Nielsen 32. Guerci B, Meyer L, Salle ´ A, et al. Comparison of S, Holman A. Cost-effectiveness of insulin aspart metabolic deterioration between insulin analog and compared to human insulin in pregnant women regular insulin after a 5-hour interruption of a with type 1 diabetes in the UK. Curr Med Res Opin. continuous subcutaneous insulin infusion in type 1 2009;25:599–605. diabetic patients. J Clin Endocrinol Metab. 1999;84:2673–8. 22. Cherubini V, Iannilli A, Iafusco D, et al. Premeal insulin treatment during basal-bolus regimen in Diabetes Ther (2018) 9:891–917 917 33. Hoogma RP, Schumicki D. Safety of insulin gluli- 42. Tubiana-Rufi N, Coutant R, Bloch J, et al. Special sine when given by continuous subcutaneous management of insulin lispro in continuous sub- infusion using an external pump in patients with cutaneous insulin infusion in young diabetic chil- type 1 diabetes. Horm Metab Res. 2006;38:429–33. dren: a randomized cross-over study. Horm Res. 2004;62:265–71. 34. Johansson UB, Adamson UC, Lins PE, Wredling RA. Improved blood glucose variability, HbA1c insu- 43. Weinzimer SA, Ternand C, Howard C, Chang CT, man Infusat and less insulin requirement in IDDM Becker DJ. Laffel LM; Insulin Aspart Pediatric Pump patients using insulin lispro in CSII. The Swedish Study Group. A randomized trial comparing con- Multicenter Lispro Insulin Study. Diabetes Metab. tinuous subcutaneous insulin infusion of insulin 2000;26:192–6. aspart versus insulin lispro in children and adoles- cents with type 1 diabetes. Diabetes Care. 35. Melki V, Renard E, Lassmann-Vague V, et al. 2008;31:210–5. Improvement of HbA1c and blood glucose stability in IDDM patients treated with lispro insulin analog 44. Institute for Quality and Efficiency in Health Care: in external pumps. Diabetes Care. 1998;21:977–82. executive summaries. Rapid-acting insulin ana- logues in children and adolescents with diabetes 36. Raskin P, Holcombe JH, Tamborlane WV, et al. A mellitus type 1-follow-up commission: executive comparison of insulin lispro and buffered regular summary of final report A08-01, version 1.0. human insulin administered via continuous sub- 2005–2009. Cologne, Germany: Institute for Qual- cutaneous insulin infusion pump. J Diabetes Com- ity and Efficiency in Health Care (IQWiG); 2009. plications. 2001;15:295–300. 45. UK National Institute for Health and Care Excel- 37. Renner R, Pfu ¨ tzner A, Trautmann M, Harzer O, lence. Diabetes (type 1 and type 2) in children and Sauter K, Landgraf R. Use of insulin lispro in con- young people: diagnosis and management. NICE tinuous subcutaneous insulin infusion treatment. guideline [NG18]. 2015. www.nice.org.uk/ Results of a multicenter trial. German Humalog- guidance/ng18 (Accessed 29 Jan 2018). CSII Study Group. Diabetes Care. 1999;22:784–8. 46. Heise T, Pieber TR, Danne T, Erichsen L, Haahr H. A 38. Schmauss S, Ko ¨ nig A, Landgraf R. Human insulin pooled analysis of clinical pharmacology trials analogue [LYS(B28), PRO(B29)]: the ideal pump investigating the pharmacokinetic and pharmaco- insulin? Diabet Med. 1998;15:247–9. dynamic characteristics of fast-acting insulin aspart in adults with type 1 diabetes. Clin Pharmacokinet. 39. Tamborlane WV, Renard E, Wadwa RP, et al. Gly- 2017;56:551–9. cemic control after 6 days of insulin pump reservoir use in type 1 diabetes: results of double-blind and 47. Fath M, Danne T, Biester T, Erichsen L, Kordonouri open-label cross-over trials of insulin lispro and O, Haahr H. Faster-acting insulin aspart provides insulin aspart. J Diabetes. 2015;7:270–8. faster onset and greater early exposure vs. insulin aspart in children and adolescents with type 1 dia- 40. van Bon AC, Bode BW, Sert-Langeron C, DeVries betes mellitus. Pediatr Diabetes. 2017;18:903–10. JH, Charpentier G. Insulin glulisine compared to insulin aspart and to insulin lispro administered by 48. Heise T, Zijlstra E, Nosek L, Rikte T, Haahr H. continuous subcutaneous insulin infusion in Pharmacological properties of faster-acting insulin patients with type 1 diabetes: a randomized con- aspart vs. insulin aspart in patients with type 1 trolled trial. Diabetes Technol Ther. diabetes receiving continuous subcutaneous insulin 2011;13:607–14. infusion: a randomized, double-blind, crossover trial. Diabetes Obes Metab. 2017;19:208–15. 41. Zinman B, Tildesley H, Chiasson JL, Tsui E, Strack T. Insulin lispro in CSII: results of a double-blind crossover study. Diabetes 1997;46:440–3. Erratum. Diabetes. 1997;46:1239.

Journal

Diabetes TherapySpringer Journals

Published: Apr 5, 2018

References

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off