EMERGING THERAPIES: DRUGS AND REGIMENS

1586 Diabetes Care Volume 44, July 2021

Switching to Once-Weekly Harpreet S. Bajaj,1,2

Richard M. Bergenstal,3
Insulin Icodec Versus Once-Daily Andreas Christoffersen,4
Melanie J. Davies,5,6 Amoolya Gowda,4
Insulin Glargine U100 in Type 2 Joakim Isendahl,4 Ildiko Lingvay,7
Peter A. Senior,8 Robert J. Silver,9
Diabetes Inadequately Controlled Roberto Trevisan,10 and
Julio Rosenstock11
on Daily Basal Insulin: A Phase 2

Downloaded from http://diabetesjournals.org/care/article-pdf/44/7/1586/633028/dc202877.pdf by guest on 17 October 2024

Randomized Controlled Trial
Diabetes Care 2021;44:1586–1594 | https://doi.org/10.2337/dc20-2877

OBJECTIVE
1
Insulin icodec (icodec) is a novel once-weekly basal insulin analog. This trial inves- LMC Diabetes and Endocrinology, Brampton,
Ontario, Canada
tigated two approaches for switching to icodec versus once-daily insulin glargine 2
Leadership Sinai Centre for Diabetes, Mount
100 units/mL (IGlar U100) in people with type 2 diabetes receiving daily basal in- Sinai Hospital, Toronto, Ontario, Canada
3
sulin and one or more oral glucose-lowering medications. International Diabetes Center at Park Nicollet,
Medical City, Minneapolis, MN
4
RESEARCH DESIGN AND METHODS Novo Nordisk A/S, Søborg, Denmark
5
Diabetes Research Centre, University of
This multicenter, open-label, treat-to-target phase 2 trial randomized (1:1:1) eligi- Leicester, Leicester, U.K.
ble basal insulin–treated (total daily dose 10–50 units) people with type 2 diabe- 6
NIHR Leicester Biomedical Research Centre,
tes (HbA1c 7.0–10.0% [53.0–85.8 mmol/mol]) to icodec with an initial 100% Leicester, U.K.
7
University of Texas Southwestern Medical
loading dose (in which only the first dose was doubled [icodec LD]), icodec with
Center, Dallas, TX
no loading dose (icodec NLD), or IGlar U100 for 16 weeks. Primary end point was 8
University of Alberta, Edmonton, Alberta, Canada
9
percent time in range (TIR; 3.9–10.0 mmol/L [70–180 mg/dL]) during weeks 15 Southern New Hampshire Diabetes and
and 16, measured using continuous glucose monitoring. Key secondary end Endocrinology, Nashua, NH
10
Azienda Socio Sanitaria Territoriale Papa
points included HbA1c, adverse events (AEs), and hypoglycemia.
Giovanni XXIII, Bergamo, Italy
11
Dallas Diabetes Research Center, Medical
RESULTS City, Dallas, TX
Estimated mean TIR during weeks 15 and 16 was 72.9% (icodec LD; n 5 54), 66.0% Corresponding author: Harpreet S. Bajaj,
(icodec NLD; n 5 50), and 65.0% (IGlar U100; n 5 50), with a statistically significant harpreet.bajaj@lmc.ca
difference favoring icodec LD versus IGlar U100 (7.9%-points [95% CI 1.8–13.9]). Received 25 November 2020 and accepted 15
Mean HbA1c reduced from 7.9% (62.8 mmol/mol) at baseline to 7.1% (54.4 mmol/ February 2021
mol icodec LD) and 7.4% (57.6 mmol/mol icodec NLD and IGlar U100); incidences Clinical trial reg. no. NCT03922750, clinicaltrials.gov
and rates of AEs and hypoglycemic episodes were comparable. This article contains supplementary material online
at https://doi.org/10.2337/figshare.14043746.
CONCLUSIONS
This article is featured in a podcast available at
Switching from daily basal insulin to once-weekly icodec was well tolerated and https://www.diabetesjournals.org/content/
provided effective glycemic control. Loading dose use when switching to once- diabetes-core-update-podcasts.
weekly icodec significantly increased percent TIR during weeks 15 and 16 versus © 2021 by the American Diabetes Association.
once-daily IGlar U100, without increasing hypoglycemia risk. Readers may use this article as long as the
work is properly cited, the use is educational
and not for profit, and the work is not altered.
More information is available at https://www.
Although many individuals with type 2 diabetes require insulin therapy at some diabetesjournals.org/content/license.
point (1,2), glycemic control remains inadequate in many patients (3). This could be See accompanying articles, pp. 1459 and 1595.
care.diabetesjournals.org Bajaj and Associates 1587

attributed to the need for frequent injec- prevent any transient deterioration in fast- Randomization
tions resulting in poor adherence and con- ing glucose levels before steady state is After screening, participants were ran-
cerns about complex dosing, the adverse achieved without jeopardizing safety, as domized (1:1:1) to once-weekly icodec
effects of hypoglycemia and weight gain, this additional dose would add to the inac- with an initial 100% loading dose (icodec
and treatment costs (2,4–8). Once-daily tive albumin-bound reservoir of icodec. LD), once-weekly insulin icodec with no
basal insulin analogs have addressed The aim of the current trial was to loading dose (icodec NLD), or once-daily
some of these concerns (9,10), but re- evaluate the effect on glycemic control IGlar U100. Randomization was per-
search suggests that people with type 2 and safety of two different approaches formed centrally using an interactive web
diabetes would value a further reduction of switching to once-weekly icodec, response system. Participants were as-
in dosing frequency of injectable therapies with and without a loading dose, versus signed to the next available treatment ac-
to once weekly, which may improve con- cording to the randomization schedule
once-daily IGlar U100 in people with
venience, adherence, and quality of life. In and stratified based on SGLT2i use (yes or
type 2 diabetes inadequately controlled
turn, these improvements may lead to no) and type of pretrial insulin therapy
on basal insulin plus at least one oral
better glycemic control (11). (i.e., receiving once-daily basal insulin, or
glucose-lowering medication.

Downloaded from http://diabetesjournals.org/care/article-pdf/44/7/1586/633028/dc202877.pdf by guest on 17 October 2024

Insulin icodec (icodec) is a novel basal recieving twice-daily basal insulin or once-
insulin analog for subcutaneous admin- daily IGlar 300 units/mL [U300]).
RESEARCH DESIGN AND METHODS
istration for the treatment of diabetes.
It has stable pharmacokinetic and phar- Research Design Procedures
macodynamic profiles, with a half-life This exploratory, multicenter, open-la- The initial dose of the trial products
of
1 week, supporting once-weekly bel, randomized, active-controlled, par- was based on the pretrial insulin daily
administration (12). The long half-life of allel-group, treat-to-target phase 2 trial doses. Participants who had been re-
icodec can be attributed to its strong, was conducted in 34 centers located in ceiving once-daily basal insulin (except
reversible albumin binding, reduced Canada, Czech Republic, Germany, Italy, for those receiving IGlar U300) at base-
enzymatic degradation, and slow recep- and the U.S. The trial consisted of a 2- line underwent a “unit to unit” switch
tor-mediated clearance (13). After sub- week screening period, 16 weeks of to icodec (700 units/mL; prefilled pen
cutaneous injection and absorption into randomized treatment, and 5 weeks of injector) (Novo Nordisk, Bagsvaerd,
the circulation, icodec monomers bind follow-up (Supplementary Fig. 1). Denmark) or IGlar U100 (100 units/mL)
to albumin to form an essentially inac- The trial was conducted in accor- (SoloSTAR prefilled pen injector; Sanofi,
tive depot, from which icodec mole- dance with the guidelines for Good Clin- Paris, France) based on their daily dose;
cules slowly reach insulin receptors at ical Practice of the International Council to derive the equivalent once-weekly
target tissues to stimulate glucose low- for Harmonisation of Technical Require- dose for the icodec LD and NLD groups,
ering. With each weekly injection, the ments for Pharmaceuticals for Human the daily dose was multiplied by 7. For
pool of albumin-bound icodec gradually Use and the Declaration of Helsinki. The those who had been receiving twice-
increases, until steady state is reached protocol, consent form, and other rele- daily basal insulin or once-daily IGlar
after 3–4 weeks when the full glucose- vant documents were approved by the U300 at baseline, starting doses of IGlar
lowering effect is achieved and insulin appropriate independent review boards U100 were decreased by 20% to mini-
clearance matches administered insulin or independent ethics committees. mize the postswitch hypoglycemia risk;
dose. At steady state, a slow, continuous for the icodec LD and NLD groups, the
release of icodec from the inactive albu- Patients
dose was similarly decreased by 20%
min-bound depot provides effective glu- The trial population comprised basal insu- and then multiplied by 7 to derive the
cose lowering throughout the week (13), lin–treated individuals with type 2 diabe- once-weekly dose. Participants in the
which was shown to be near evenly dis- icodec LD group received an initial
tes recruited between 9 May 2019 and
tributed across a 1-week dosing interval 100% loading dose with the first dose
15 August 2019 with the last follow-up
(12). In a phase 2 trial in insulin-naive of icodec (i.e., the first weekly dose was
occurring on 27 January 2020. Eligible
individuals with type 2 diabetes, once- doubled), and then it was reverted to
participants, aged 18–75 years with a
weekly icodec resulted in a similar glu- the calculated weekly dose at week 2
BMI of #40 kg/m2 and a centrally as-
cose-lowering effect and safety profile (Supplementary Fig. 1B). Icodec had to
sessed HbA1c of 7.0–10.0% (53.0–85.8 be taken subcutaneously once weekly
compared with once-daily insulin glargine
100 units/mL (IGlar U100) (14). mmol/mol), were treated for at least 90 on the same day each week, while IGlar
Theoretically, individuals switching from days before screening with stable doses U100 had to be taken subcutaneously
a daily basal insulin regimen may require of metformin, with or without concomi- once daily at any time of the day but at
additional insulin to maintain glycemic con- tant use of dipeptidyl peptidase 4 inhibi- the same time every day.
trol during the initial weeks after switch, tors and/or sodium–glucose cotransporter A treat-to-target approach was used
until steady state is reached. Pharmacoki- 2 inhibitors (SGLT2is), along with basal in- to ensure optimal titration of insulin;
netic and pharmacodynamic modeling sulin (once or twice daily) with a total dai- doses of icodec and IGlar U100 were ti-
data, based on clinical pharmacology stud- ly dose of 10–50 units. Full details of trated weekly to achieve a prebreakfast
ies of icodec, suggest that when switching inclusion and exclusion criteria are provid- self-measured blood glucose (SMBG) tar-
from daily to once-weekly basal insulin, ed in Supplementary Table 1. All partici- get of 4.4–7.2 mmol/L (80–130 mg/dL).
adding a supplemental dose of icodec (a pants provided written informed consent Participants measured SMBG levels (Free-
“loading dose”) to the first dose may to participate in the trial. Style Precision/Optium Neo Meter; Abbott
1588 Switch to Once-Weekly Icodec in Type 2 Diabetes Diabetes Care Volume 44, July 2021

GmbH & Co. KG, Abbott Diabetes Care) Statistical Analysis methods were used for imputing inter-
each day before breakfast. Doses were The sample size was chosen as the mittent missing values. Because this is a
uptitrated weekly by 4 units/day (IGlar number of patients required for the phase 2 trial and it is exploratory in na-
U100) or 28 units/week (icodec LD 95% CI for TIR 3.9–10.0 mmol/L (70–80 ture, no adjustments were made for
and icodec NLD) if the mean of the mg/dL) of any pairwise comparison hav- multiplicity. The number of hypoglycemic
three most recent prebreakfast SMBG ing a width of 2.5 h/24 h (corresponding events was analyzed using a negative bi-
values in a week was >7.2 mmol/L to a difference of 10%-points), with a nomial regression model with log link.
(>130 mg/dL). However, if any one of probability of 80%, based on an as- The model included treatment, pretrial
the three prebreakfast SMBG values sumed SD of 3.0 h/24 h (corresponding insulin treatment, and SGLT2i use as
was <4.4 mmol/L (<80 mg/dL), the insu- to 12%); based on this, 50 participants fixed factors and the logarithm of the
lin doses were downtitrated 4 units/day per treatment arm were required. This time period for which the hypoglycemic
(IGlar U100) or 28 units/week (icodec LD trial was exploratory in nature and in- episodes were considered as an offset.
and icodec NLD). tended to inform the design of the The on-treatment period was defined
Background oral glucose-lowering drugs phase 3 program; therefore, the power as the period from the date of first trial

Downloaded from http://diabetesjournals.org/care/article-pdf/44/7/1586/633028/dc202877.pdf by guest on 17 October 2024

were continued unchanged throughout was not calculated. drug dose until the last follow-up visit
the trial unless there were safety concerns TIR during weeks 15 and 16 for each or the last dosing day of randomized
in the opinion of the investigator. individual was calculated as the number treatment plus 5 weeks (for IGlar U100)
of recorded measurements in the range or 6 weeks (for icodec), whichever
3.9–10.0 mmol/L (70–180 mg/dL) divid- came first, and represents the time peri-
Outcomes
Participants wore a continuous glucose ed by the total number of recorded od during which participants are consid-
measurements over 14 days, multiplied ered to be exposed to the trial product.
monitoring (CGM) system (Dexcom G6;
by 100. The full analysis set consisted of all ran-
Dexcom, Inc., San Diego, CA) through-
The primary estimand (trial product domized participants. The safety analysis
out the screening and treatment peri-
estimand) was defined as the mean dif- set consisted of all participants exposed
ods; sensors were changed weekly.
ference in the primary end point be- to at least one dose of trial product. Data
CGM data were blinded for both partici-
tween each icodec group and IGlar were analyzed using SAS software 9.4. No
pants and investigators and were not
U100 for all randomized participants, if data monitoring committee oversaw the
used for insulin dose titration or report-
they had adhered to the randomized in- trial. This trial was registered on the Clini-
ing of hypoglycemic episodes. Partici-
sulin treatment without initiation of res- calTrials.gov website (NCT03922750).
pants were asked to measure SMBG if
cue medication (initiation of insulin
they experienced symptoms suggestive
treatment other than the randomized
of hypoglycemia at any time of the day. Data and Resource Availability
treatment) and had recorded $70% of
The primary end point was percent The data sets generated during and/or
the planned CGM measurements during
time in range (TIR; 3.9–10.0 mmol/L analyzed during the current trial are
weeks 15 and 16. A more detailed expla-
[70–180 mg/dL]), measured by CGM, dur- available from the corresponding author
nation is provided in the Supplementary
ing weeks 15 and 16 of treatment. Sup- on reasonable request.
Material. For TIR, the response throughout
portive secondary efficacy end points the last 2 weeks of treatment (weeks 15
were changes in HbA1c, fasting plasma RESULTS
and 16) was analyzed using an ANCOVA
glucose (FPG), and body weight from model with randomized treatment, pretrial Of 222 people screened, 154 individuals
baseline to week 16 and weekly insulin insulin treatment, and SGLT2i use as fixed were randomized (icodec LD, n 5 54; ico-
doses during weeks 15 and 16 of treat- factors and baseline TIR as covariate. Miss- dec NLD, n 5 50; and IGlar U100, n 5 50),
ment. Safety end points were number of ing end point values were imputed by and 152 completed the 16-week treatment
on-treatment adverse events (AEs) from multiple imputation using data from partic- period (Supplementary Fig. 2). All 154 trial
baseline to week 21 and number of ipants in the same randomized group participants were included in the efficacy
self-reported hypoglycemic events docu- based on an ANCOVA model with baseline and safety analyses. Eight participants had
mented by SMBG or by requirement for TIR as a covariate. Each imputed data set missing or <70% CGM measurements dur-
external assistance for recovery. Hypogly- was analyzed separately, and estimates ing the last 2 weeks of treatment (weeks
cemic events were classified as: level 1 were then combined using Rubin’s rule 15 and 16) and required imputation of
(blood glucose <3.9 mmol/L and $3.0 (16). Supportive secondary efficacy end data for the primary end point: one for ico-
mmol/L [<70 mg/dL and $54 mg/dL]); a points were analyzed in the same way as dec LD, five for icodec NLD, and two for
combination of level 2 (blood glucose the primary end point, except for the IGlar U100. Three participants started insu-
<3.0 mmol/L [<54 mg/dL]) and level 3 mean weekly insulin dose during the last 2 lin aspart as a rescue medication during
(hypoglycemia with severe cognitive im- weeks of treatment, which was log-trans- the trial; two participants (one each for ico-
pairment requiring external assistance for formed and analyzed using an ANOVA dec NLD and IGlar U100) started insulin as-
recovery); or level 3 alone (15). AEs of model. Missing data for secondary end part on the same day or after the last
special interest were major cardiovascular points were imputed based on data dose of trial drug, and one participant
AEs, hypersensitivity and injection-site re- from participants in the same random- (icodec NLD) started insulin aspart on day
actions, and death; these were reviewed ized group using a sequential conditional 30. Separately, one participant (IGlar U100)
by a blinded independent adjudication regression approach including all postba- discontinued sitagliptin/metformin treat-
committee. seline values; Markov chain Monte-Carlo ment during the trial.
care.diabetesjournals.org Bajaj and Associates 1589

Baseline demographics and clinical charac- P 5 0.01) and similar between the ico- 25.0%, and 30.6% for the icodec LD, ico-
teristics were generally similar among groups dec NLD group and the IGlar U100 group dec NLD, and IGlar U100 groups, respectively
(Table 1 and Supplementary Table 2): most (ETD, 1.01%-points [95% CI –5.33 to 7.5]; (no statistical analyses were performed).
participants were male (72.1%); the mean P 5 nonsignificant [NS]) (Fig. 1). Compared with baseline, FPG was
age was 61.7 years; the mean duration of HbA1c levels reduced in all treatment lower at week 16 in all treatment
type 2 diabetes was 15.1 years; the mean ± groups during the trial (Fig. 2A). At groups; estimated mean values at week
SD HbA1c was 7.9% ± 0.7 (62.8 ± 7.7 mmol/ week 16, estimated mean HbA1c levels 16 were 7.3 mmol/L (132 mg/dL), 7.2
mol); the mean ± SD FPG was 8.0 ± 2.1 were 7.1% (54.4 mmol/mol), 7.4% (57.6 mmol/L (129 mg/dL), and 7.4 mmol/L
mmol/L (144 ± 37 mg/dL); and the majority mmol/mol), and 7.4% (56.9 mmol/mol) (134 mg/dL) for icodec LD, icodec NLD,
of participants (57.8%) were taking more in the icodec LD, icodec NLD, and IGlar and IGlar U100, respectively (Fig. 2B).
than one oral glucose-lowering drug. IGlar U100 groups, respectively. The estimat- The estimated mean change from base-
U100 was the most used basal insulin at ed mean change from baseline at week line was similar between the icodec LD
screening (59.7%). 16 was –0.8%-points (–8.4 mmol/mol) and IGlar U100 groups (–0.7 mmol/L vs.
The estimated mean TIR during weeks in the icodec LD group, –0.5%-points –0.6 mmol/L: ETD, –0.12 mmol/L [95%

Downloaded from http://diabetesjournals.org/care/article-pdf/44/7/1586/633028/dc202877.pdf by guest on 17 October 2024

15 and 16 was 72.9% for icodec LD, 66.0% (–5.2 mmol/mol) in the icodec NLD CI –0.74 to 0.50], P 5 NS [–12 mg/dL
for icodec NLD, and 65.0% for IGlar group, and –0.5%-points (–5.9 mmol/mol) vs. –10 mg/dL: ETD, –2.20 mg/dL (95%
U100, representing estimated changes from in the IGlar U100 group, with no statisti- CI –13.33 to 8.94), P 5 NS]) and also
baseline in TIR of 15.4%-points, 8.6%- cally significant differences between the between the icodec NLD and IGlar U100
points, and 7.6%-points for icodec LD, ico- icodec groups and IGlar U100 group (ETD, groups (–0.8 mmol/L vs. –0.6 mmol/L:
dec NLD, and IGlar U100, respectively. icodec LD vs. IGlar U100, –0.23%-points ETD, –0.26 mmol/L [95% CI –0.90 to
These correspond to estimated changes [95% CI –0.49 to 0.02] [–2.53 mmol/mol 0.38], P 5 NS [–15 mg/dL vs. –10 mg/
from baseline in TIR of
3 h 42 min/day, 2 (95% CI –5.33 to 0.27)], P 5 NS; ETD, ico- dL: ETD, –4.74 mg/dL (95% CI –16.28 to
h 4 min/day, and 1 h 49 min/day, respec- dec NLD vs. IGlar U100, 0.07%-points 6.80), P 5 NS]). At week 8, mean ob-
tively. The primary end point of TIR was sta- [95% CI –0.19 to 0.33] [0.73 mmol/mol served FPG values were similar between
tistically significantly greater for the icodec (95% CI –2.11 to 3.57)], P 5 NS). After 16 the icodec LD and IGlar U100 groups
LD group compared with the IGlar U100 weeks of treatment, the proportions of (7.4 mmol/L vs. 7.3 mmol/L [133 mg/dL
group (estimated treatment difference participants achieving a target HbA1c of vs. 132 mg/dL]) but slightly higher for
[ETD], 7.88%-points [95% CI 1.83–13.93]; <7.0% (53 mmol/mol) were 44.2%, the icodec NLD group (8.2 mmol/L

Table 1—Baseline demographics and clinical characteristics

Icodec LD (n = 54) Icodec NLD (n = 50) IGlar U100 (n = 50) Total (N = 154)
Age, years 62.4 ± 7.2 62.1 ± 8.2 60.5 ± 7.9 61.7 ± 7.8
Male, n (%) 39 (72.2) 39 (78.0) 33 (66.0) 111 (72.1)
Duration of type 2 diabetes, years 13.8 ± 7.7 16.8 ± 8.2 14.8 ± 8.1 15.1 ± 8.1
Ethnicity, n (%)
White 46 (85.2) 39 (78.0) 44 (88.0) 129 (83.8)
Asian 4 (7.4) 9 (18.0) 3 (6.0) 16 (10.4)
Black or African American 3 (5.6) 2 (4.0) 3 (6.0) 8 (5.2)
Native Hawaiian or other Pacific Islander 1 (1.9) 0 0 1 (0.6)
BMI, kg/m2 30.2 ± 4.3 29.0 ± 4.1 30.3 ± 5.0 29.8 ± 4.5
FPG, mmol/L 7.9 (1.9)* 8.0 (2.3) 8.2 (2.0)† 8.0 (2.1)
FPG, mg/dL 142 ± 34* 144 ± 41 148 ± 36† 144 ± 37
HbA1c, % 7.8 (0.7) 7.9 (0.7) 7.9 (0.7) 7.9 (0.7)
HbA1c, mmol/mol 62.0 (7.4) 63.3 (8.0) 63.2 (7.8) 62.8 (7.7)
TIR at baseline, % 58.9 (23.2) 54.5 (20.2) 58.7 (21.5) 57.4 (21.7)
Total insulin dose‡ (CV%), units 22.5 (61.0) 24.5 (47.7) 24.0 (49.2) 23.6 (52.8)
Basal insulin at screening, n (%)
Insulin degludec 4 (7.4) 15 (30.0) 7 (14.0) 26 (16.9)
Insulin detemir 9 (16.7) 3 (6.0) 1 (2.0) 13 (8.4)
Insulin glargine U100 32 (59.3) 23 (46.0) 37 (74.0) 92 (59.7)
Insulin glargine U300 9 (16.7) 8 (16.0) 5 (10.0) 22 (14.3)
NPH (isophane) insulin 0 1 (2.0) 0 1 (0.6)
Data are mean (SD) unless otherwise indicated. All percentages are subject to rounding. CV%, coefficient of variation. *n = 52. †n = 49.
‡Geometric mean.
1590 Switch to Once-Weekly Icodec in Type 2 Diabetes Diabetes Care Volume 44, July 2021

frequent AEs are provided in

Supplementary Table 3.
Two participants reported seven seri-
ous AEs in the icodec LD group (including
one event of acute myocardial infarction),
and one participant reported one serious
AE (acute myocardial infarction) in the
IGlar U100 group (Table 2). Both acute
myocardial infarction events were con-
firmed by adjudication and assessed as
unlikely to be related to the trial prod-
ucts. The participant in the icodec LD
group who experienced acute myocardial
infarction was withdrawn from the trial;

Downloaded from http://diabetesjournals.org/care/article-pdf/44/7/1586/633028/dc202877.pdf by guest on 17 October 2024

no other discontinuations due to AEs oc-
curred. None of the AEs led to a reduc-
tion in the dose of trial medication.
No deaths were reported during the
Figure 1—TIR during the last 2 weeks of the treatment period (full analysis set). TIR was the pri- trial. Two participants (4.0%) in the ico-
mary end point. TAR, time above range; TBR, time below range.
dec NLD group had hypersensitivity reac-
tions confirmed by adjudication, which
[144 mg/dL]). No statistical analyses U100 groups (ETD, 0.51 kg [95% CI –0.44 were reported as not being related to
at week 8 were performed. to 1.47]; P 5 NS), a statistically signifi- the trial treatment (one participant had
For comparison, the weekly changes cantly greater increase in body weight a skin rash with irritation attributed to
in fasting (prebreakfast) SMBG levels was seen with icodec NLD compared an adhesive bandage and another had a
(measured as part of the titration pro- with IGlar U100 (ETD, 1.22 kg [95% CI skin rash at the CGM insertion site). All
cess and not as an end point) are shown 0.24–2.2]; P 5 0.01). injection-site reactions reported in the
in Fig. 2C. SMBG levels decreased in all Hypoglycemic events over 21 weeks trial were mild (Table 2).
treatment groups over the course of the (on-treatment period) are summarized
trial; however, in the icodec NLD group, in Table 2. The percentage of partici- CONCLUSIONS
mean SMBG levels were seen to increase pants with one or more level 2 or level The results of the current trial suggest
slightly versus baseline during the first 3 3 hypoglycemic events during this period that, in people with type 2 diabetes re-
weeks before reverting to levels similar was 7.4%, 4.0%, and 12.0% in the icodec ceiving daily basal insulin therapy, switch-
to those for the icodec LD and IGlar LD, icodec NLD, and IGlar U100 groups, ing to once-weekly icodec is effective and
U100 groups at about week 5. respectively, with rates per patient-year well tolerated. Compared with switching
Insulin dose gradually increased dur- of exposure of 0.78, 0.15, and 0.79, re- to once-daily basal insulin IGlar U100,
ing the trial in all groups (Fig. 2D). Dur- spectively. No level 3 (severe) hypoglyce- switching to once-weekly icodec displayed
ing the last 2 weeks of treatment mic events were reported in the trial. similar or greater efficacy regarding the
(weeks 15 and 16), the estimated mean Similar results were observed for the peri- primary end point of TIR during weeks 15
weekly doses were 191 units (2.18 od from baseline to week 16 (data not and 16. Switching to icodec using an ini-
units/kg), 242 units (2.81 units/kg) and shown). The rate and pattern of level 1 tial 100% LD (doubling of the first dose)
196 units (2.27 units/kg) for icodec LD, hypoglycemia (Fig. 2E) over the on-treat- resulted in statistically significantly longer
icodec NLD, and IGlar U100, respectively ment period were similar across all TIR than switching to IGlar U100. Further-
(estimated treatment ratio, icodec LD vs. groups. The rate and pattern of combined more, doubling the initial dose did not in-
IGlar U100, 0.98 units [95% CI 0.78–1.23], level 2 and level 3 hypoglycemic episodes crease the risk of hypoglycemia at any
P 5 NS [0.96 units/kg (95% CI 0.78–1.19), (Fig. 2F) were similar between the icodec point during the trial period and pre-
P 5 NS]; estimated treatment ratio, ico- LD and IGlar U100 groups and appeared vented the mild, transient prebreakfast
dec NLD vs. IGlar U100, 1.24 units [95% to be lower for the icodec NLD group SMBG elevation observed in those ran-
CI 0.98–1.56], P 5 NS [1.24 units/kg (95% than for the IGlar U100 group. No initial domized to icodec without a loading dose
CI 1.00–1.54), P 5 0.047]). increase in level 1 or combined level 2 during the immediate postswitch period;
The change in body weight during and level 3 hypoglycemia was observed that could lead clinicians or patients to in-
the trial is shown in Supplementary in the icodec LD group. advertently perceive that the once-weekly
Fig. 3. Estimated mean changes from As summarized in Table 2, the inci- insulin may not be as effective as the pre-
baseline were 0.6 kg, 1.3 kg, and 0.1 kg dence and rates of AEs were similar vious basal insulin.
in the icodec LD, icodec NLD, and IGlar across treatment groups. Most AEs Notably, a greater change from base-
U100 groups, respectively. Although the were mild in severity, and only a small line in TIR (15.4%-points, corresponding
mean change in body weight from base- number of AEs in each group were pos- to
3 h 42 min/day) was observed dur-
line was not statistically significantly dif- sibly or probably related to treatment, ing weeks 15 and 16 in the icodec LD
ferent between the icodec LD and IGlar none of which were serious. The most group (vs. 8.6%-points [2 h 4 min/day]
care.diabetesjournals.org Bajaj and Associates 1591

Downloaded from http://diabetesjournals.org/care/article-pdf/44/7/1586/633028/dc202877.pdf by guest on 17 October 2024

Figure 2—Key parameters during the trial. A: Mean change in HbA1c from baseline to week 16 (FAS). B: Mean change in FPG from baseline to week
16 (FAS). C: Mean prebreakfast SMBG levels over time (FAS). D: Mean weekly insulin dose over time (FAS). E: Cumulative number of level 1 (“alert”
value) hypoglycemic episodes per patient (SAS). F: Cumulative number of level 2 (clinically significant) or level 3 (severe) hypoglycemic episodes
per patient (SAS). Observed data. A–C: Mean ± SEM. D: Geometric mean ± SEM on log-scale back transformed. For C, SMBG was assessed with a
blood glucose meter as plasma equivalent of capillary whole blood glucose. For D, the dose for a given visit represents the total dose during the
preceding week, and weekly IGlar U100 doses were derived as seven times the average daily dose during the preceding week. *Estimated mean
values and the corresponding CI at week 16 derived based on multiple imputation. FAS, full analysis set; SAS, safety analysis set; U, unit.

in the icodec NLD group and 7.6%- considered a clinically significant improve- people with type 2 diabetes (17–22).
points [1 h 49 min/day] in the IGlar U100 ment in glycemic control (17) and that Moreover, it is important to note that a
group); this is an important observation TIR improvement has been associated mean TIR of >70%, a target recom-
given that a 5% increase in TIR is with clinically significant benefits for mended by the International Consensus
Table 2—On-treatment hypoglycemic episodes and AEs, including AEs of special interest (SAS; N = 154)
RR (95% CI)* RR (95% CI)*
Icodec LD (n = 54, 21.8 PYE) Icodec NLD (n = 50, 20.3 PYE) IGlar U100 (n = 50, 20.2 PYE) icodec LD/IGlar icodec NLD/IGlar
n (%) Events Rate* n (%) Events Rate n (%) Events Rate U100 U100
Hypoglycemic episodes
Hypoglycemia “alert” value (level 1)† 19 (35.2) 83 3.809 26 (52.0) 87 4.29 22 (44.0) 76 3.77 0.94 (0.44–1.98) 1.10 (0.51–2.35)
Clinically significant (level 2)‡ or severe 4 (7.4) 17 0.78 2 (4.0) 3 0.15 6 (12.0) 16 0.79 0.87 (0.18–4.24) 0.39 (0.05–2.80)
(level 3) hypoglycemia
Severe hypoglycemia (level 3)§ 0 0 0
AEs
1592 Switch to Once-Weekly Icodec in Type 2 Diabetes

All AEs 28 (51.9) 85 3.90 30 (60.0) 77 3.80 23 (46.0) 76 3.77

Serious AEs 2 (3.7) 7k 32.1 0 1 (2.0) 1¶ 0.05
Fatal AEs 0 0 0
Severe AEs 1 (1.9) 1 4.6 1 (2.0) 1 0.05 0
AEs leading to withdrawal 1 (1.9) 1 4.6 0 0
AEs probably related to basal insulin 1 (1.9) 2 9.2 2 (4.0) 5 0.247 3 (6.0) 3 0.15
AEs possibly related to basal insulin 0 1 (2.0) 1 0.05 2 (4.0) 2 0.10
AEs of special interest
Injection-site reaction# 1 (1.9) 2 9.2 1 (2.0) 4 0.20 2 (4.0) 3 0.15
Hypersensitivity event# 2 (4.0) 2 9.9
MACE# 1 (1.9) 1k** 4.6 0 1 (2.0) 1¶** 0.05
The on-treatment period represents the time period during which participants are considered to be exposed to the trial product. MACE, major adverse cardiovascular event; PYE, patient-years of expo-
sure; RR, rate ratio; SAS, safety analysis set. *The number of hypoglycemic events was analyzed post hoc using a negative binomial regression model with log link. The model included treatment, pre-
trial insulin treatment, and SGLT2i use as fixed factors and the logarithm of the time period for which the hypoglycemic episodes were considered as an offset. *Rate: number of events per patient-
year of exposure. †Hypoglycemia “alert” value (level 1): blood glucose value <3.9 mmol/L (<70 mg/dL) and $3.0 mmol/L ($54 mg/dL) confirmed by blood glucose meter. ‡Clinically significant hypogly-
cemia (level 2): blood glucose value <3.0 mmol/L (<54 mg/dL). §Severe hypoglycemia (level 3): hypoglycemia with severe cognitive impairment requiring external assistance for recovery. kAcute myo-
cardial infarction (confirmed by adjudication), fall, joint injury, and upper limb and two facial bone fractures in one patient; muscle abscess in one patient. ¶Myocardial infarction (confirmed by
adjudication). #Independently adjudicated and confirmed. **Also included as serious AEs.
Diabetes Care Volume 44, July 2021

Downloaded from http://diabetesjournals.org/care/article-pdf/44/7/1586/633028/dc202877.pdf by guest on 17 October 2024

care.diabetesjournals.org Bajaj and Associates 1593

on Time in Range (17), was achieved in on Time in Range across all treatment associated with improved treatment satis-
the icodec LD group during the last 2 groups. faction and medication-taking behavior
weeks of treatment. Notably, despite the Results for the other safety parame- (23–25). In type 2 diabetes, medication
short trial duration limiting the time pa- ters measured also show that icodec LD adherence and timely insulin initiation are
tients had to reach target, the HbA1c re- was well tolerated. Most AEs across critical for maintaining glycemic control
ductions and proportion of participants treatment groups were mild in severity, and reducing diabetes-related complications
who achieved HbA1c <7.0% (<53.0 and most (including the small number and associated health care resource utiliza-
mmol/mol) with icodec are consistent of serious AEs) were assessed to be un- tion (26–30).
with the observed increase in percent related to treatment. None of the AEs Overall, the safety and glycemic efficacy
TIR. There was a statistically significant required a reduction in the dose of trial of icodec seen in this first phase 2 trial
but small increase in body weight at the medication, and there were few AEs of with icodec in basal insulin–treated partici-
end of the trial in the icodec NLD group special interest. pants reflect the findings of two completed
compared with the IGlar U100 group; this The strengths of this trial include its treat-to-target trials with icodec in insulin-
may reflect the numerically higher insulin randomized, multicenter design and the naive people with type 2 diabetes (Clinical-

Downloaded from http://diabetesjournals.org/care/article-pdf/44/7/1586/633028/dc202877.pdf by guest on 17 October 2024

dose requirement in the former. low treatment discontinuation rate. Addi- Trials.gov identifiers NCT03751657 and
In this trial, administration of a load- tionally, this trial used CGM and measured NCT03951805) (14). Taken together, these
ing dose of icodec when switching from TIR as a primary end point in a patient phase 2 trial results suggest effective glyce-
daily basal insulin seemed to have pre- population with type 2 diabetes, which is mic control and comparable safety of ico-
vented the mild and transient elevation quite unique for comparative insulin stud- dec versus IGlar U100 in a broad patient
of prebreakfast SMBG levels seen in ies. TIR is a clinically recognized parameter, population with type 2 diabetes, with the
the icodec NLD group. This may be ex- and internationally accepted TIR targets major clinical value of reducing the number
plained by the molecular characteristics now exist to guide diabetes management of weekly basal insulin injections from sev-
and pharmacokinetic and pharmacody- (15,17,19). Limitations of the trial include en to only one.
namic properties of icodec. While the its relatively short duration and the mod- In conclusion, in a patient population
long half-life of icodec makes it suitable est sample size. In addition, the trial used with type 2 diabetes receiving daily bas-
an open-label design; however, the CGM al insulin therapy, the use of a loading
for once-weekly dosing, it also extends the
recordings used to analyze the primary dose when switching to once-weekly
time to steady state (by
3–4 weeks). This
end point of TIR were blinded to both the icodec resulted in effective glycemic
could consequently produce an initial mild
investigator and trial participant. The ob- control without a transient elevation of
deterioration in fasting glucose levels in
served differences in TIR at weeks 15 and fasting glucose levels during the switch
those switching from a prior basal insulin
16 could be due to minor differences in and without increasing the risk of clini-
therapy who will have an initial treatment
patient characteristics between treatment cally relevant hypoglycemia compared
deficit. Pharmacokinetic and pharmacody-
groups and the low patient numbers con- with IGlar U100. These results will be
namic modeling indicated that using a
cerned. Although individuals taking met- considered when switching to once-
loading dose of icodec could offset this by
formin and/or dipeptidyl peptidase 4 weekly insulin icodec in the phase 3
ensuring the availability of an adequate al- inhibitors and/or SGLT2is were eligible
bumin-bound inactive reservoir; slow re- clinical development program.
to participate, those who were taking oth-
lease of icodec molecules would provide er glucose-lowering agents and patients
effective glucose lowering without an in- with certain comorbidities were precluded Acknowledgements. The authors thank Dr.
creased risk of hypoglycemia versus once- from participation, limiting the generaliz- Gemma Rogers of Oxford PharmaGenesis (Ox-
daily basal insulin. This hypothesis is sup- ability of the results. ford, U.K.) (supported by Novo Nordisk) for
ported by the results of the current trial, The potential to administer basal insulin providing writing assistance.
because the rate and pattern of hypogly- Duality of Interest. H.S.B. has received speak-
once weekly, thereby reducing the num- ing fees from AstraZeneca, Eli Lilly and Company,
cemic episodes, irrespective of level, were ber of injections, has important implica- Janssen Pharmaceuticals, Merck, and Novo Nor-
similar between the icodec LD and IGlar tions for diabetes management. Research disk and research funding paid to LMC Health-
U100 groups both immediately postswitch shows that people with diabetes would care for serving as principal investigator on
and throughout the treatment period. welcome such a reduction to improve clinical trials from Amgen, AstraZeneca, Boeh-
The results from this trial suggest ringer Ingelheim, Ceapro Inc., Eli Lilly and Compa-
convenience and quality of life; conse- ny, Gilead Sciences, Inc., Janssen Pharmaceuticals,
that switching from an existing basal in- quently, it may help to improve adherence Kowa Pharmaceuticals Co. Ltd., Madrigal Pharma-
sulin to icodec, with or without a load- and overall glycemic control further (11). ceuticals, Merck, Novo Nordisk, Sanofi, and Trici-
ing dose, provides effective glycemic In a study by Peyrot et al. (6), 27.6% of da, Inc. R.M.B. has received research support
control with comparable risk of hypogly- people with diabetes (n 5 1,530) de- from, consulted for, or has been on a scientific
advisory board for Abbott Diabetes Care, As-
cemia. It was reassuring that not a sin- scribed “taking insulin at prescribed time censia, Dexcom, Inc., Eli Lilly and Company,
gle episode of severe hypoglycemia or with meals every day” as difficult, and Hygieia Biological Laboratories, Johnson &
(level 3) was reported for any treatment 92.5% of them expressed a wish for good Johnson, Medtronic, Novo Nordisk, Onduo,
group throughout the trial duration and control with “insulin [that did] not [need Roche, Sanofi, and UnitedHealthcare (R.M.B.’s
that the time spent below range (<3.9 to be] injected every day.” Data from employer, nonprofit HealthPartners Institute,
contracts for his services, and no personal in-
mmol/L [<54 mg/dL]) during weeks 15 studies evaluating glucagon-like peptide 1 come goes to him). M.J.D. has acted as a con-
and 16 was well below the 4% target rec- receptor agonists demonstrate that once- sultant, advisory board member, and speaker
ommended by the International Consensus weekly administration of these agents is for AstraZeneca, Boehringer Ingelheim, Janssen
1594 Switch to Once-Weekly Icodec in Type 2 Diabetes Diabetes Care Volume 44, July 2021

Pharmaceuticals, Eli Lilly and Company, Merck of Medical Care in Diabetes—2020 [published 17. Battelino T, Danne T, Bergenstal RM, et al.
Sharp & Dohme, Novo Nordisk, and Sanofi; correction appears in Diabetes Care 2020; Clinical targets for continuous glucose monitoring
served as an advisory board member for Gilead 43:1979]. Diabetes Care 2020;43(Suppl. 1): data interpretation: recommendations from the
Sciences, Inc. and Servier; served as a speaker S98–S110 international consensus on time in range. Diabetes
for Mitsubishi Tanabe Pharma Corporation, Napp 2. Nyenwe EA, Jerkins TW, Umpierrez GE, Care 2019;42:1593–1603
Pharmaceuticals, and Takeda Pharmaceuticals In- Kitabchi AE. Management of type 2 diabetes: 18. Beck RW, Bergenstal RM, Cheng P, et al. The
ternational Inc.; and has received grants in sup- evolving strategies for the treatment of patients relationships between time in range, hyperglycemia
port of investigator and investigator-initiated with type 2 diabetes. Metabolism 2011;60:1–23 metrics, and HbA1c. J Diabetes Sci Technol 2019;13:
trials from AstraZeneca, Boehringer Ingelheim, 3. Ross SA, Tildesley HD, Ashkenas J. Barriers to 614–626
Janssen Pharmaceuticals, Eli Lilly and Company, effective insulin treatment: the persistence of 19. Beck RW, Bergenstal RM, Riddlesworth TD,
Novo Nordisk, and Sanofi. I.L. received research poor glycemic control in type 2 diabetes. Curr et al. Validation of time in range as an outcome
funding, advisory/consulting fees, and/or other Med Res Opin 2011;27(Suppl. 3):13–20 measure for diabetes clinical trials. Diabetes Care
support from AstraZeneca, Boehringer Ingelheim, 4. Escalada J, Orozco-Beltran D, Morillas C, et al. 2019;42:400–405
Eli Lilly and Company, GI Dynamics, Intarcia Ther- Attitudes towards insulin initiation in type 2 20. Lu J, Ma X, Shen Y, et al. Time in range is
apeutics, Intercept Pharmaceuticals, Janssen diabetes patients among healthcare providers: a associated with carotid intima-media thickness in
Pharmaceuticals, MannKind Corporation, Merck, survey research. Diabetes Res Clin Pract 2016; type 2 diabetes. Diabetes Technol Ther 2020;22:
Mylan, Novartis, Novo Nordisk, Pfizer, Sanofi, Tar- 122:46–53 72–78

Downloaded from http://diabetesjournals.org/care/article-pdf/44/7/1586/633028/dc202877.pdf by guest on 17 October 2024

get PharmaSolutions, Inc., Valeritas, Inc., and 5. Karter AJ, Subramanian U, Saha C, et al. 21. Lu J, Ma X, Zhou J, et al. Association of time
Zealand Pharma. P.A.S. received research support Barriers to insulin initiation: the translating in range, as assessed by continuous glucose
to his institution from Allergan, Novo Nordisk, research into action for diabetes insulin starts monitoring, with diabetic retinopathy in type 2
and ViaCyte; as chair of Diabetes Canada Clinical
project. Diabetes Care 2010;33:733–735 diabetes. Diabetes Care 2018;41:2370–2376
Practice Guidelines, he does not provide con-
6. Peyrot M, Barnett AH, Meneghini LF, Schumm- 22. Vigersky RA, McMahon C. The relationship of
sulting or speaking services to, and receives
Draeger PM. Insulin adherence behaviours and hemoglobin A1C to time-in-range in patients
no personal fees from, industry partners. R.J.S.
barriers in the multinational Global Attitudes of with diabetes. Diabetes Technol Ther 2019;21:
serves on speakers bureaus for AstraZeneca
Patients and Physicians in Insulin Therapy study. 81–85
and Novo Nordisk and serves on advisory pan-
Diabet Med 2012;29:682–689 23. Drucker DJ, Buse JB, Taylor K, et al.;
els for Novo Nordisk. R.T. received lecture fees
7. Okemah J, Peng J, Qui~ nones M. Addressing DURATION-1 Study Group. Exenatide once weekly
from AstraZeneca, Boehringer Ingelheim, Eli
clinical inertia in type 2 diabetes mellitus: a versus twice daily for the treatment of type 2
Lilly and Company, Novo Nordisk, and Sanofi
and has participated in advisory panels for As- review. Adv Ther 2018;35:1735–1745 diabetes: a randomised, open-label, non-inferiority
traZeneca, Boehringer Ingelheim, Eli Lilly and 8. Donnelly LA, Morris AD; DARTS/MEMO study. Lancet 2008;372:1240–1250
Company, Novo Nordisk, and Sanofi. J.R. has collaboration. Adherence to insulin and its 24. Nguyen H, Dufour R, Caldwell-Tarr A.
participated in advisory panels for Applied Ther- association with glycaemic control in patients Glucagon-like peptide-1 receptor agonist (GLP-
apeutics, Boehringer Ingelheim, Eli Lilly and with type 2 diabetes. QJM 2007;100:345–350 1RA) therapy adherence for patients with type 2
Company, Intarcia Therapeutics, Janssen Phar- 9. Cheng AYY, Patel DK, Reid TS, Wyne K. diabetes in a medicare population. Adv Ther
maceuticals, Novartis, Novo Nordisk, Oramed Differentiating basal insulin preparations: 2017;34:658–673
Pharmaceuticals, and Sanofi and has received understanding how they work explains why they 25. Takase T, Nakamura A, Yamamoto C, et al.
research support from Eli Lilly and Company, are different. Adv Ther 2019;36:1018–1030 Improvement in treatment satisfaction after
Genentech, GlaxoSmithKline, Intarcia Therapeu- 10. Hernando VU, Pablo FJ. Efficacy and safety switching from liraglutide to dulaglutide in patients
tics, Janssen Pharmaceuticals, Merck & Co., Novo of the second generation basal insulin analogs in with type 2 diabetes: a randomized controlled trial.
Nordisk, Oramed Pharmaceuticals, Pfizer, and Sa- type 2 diabetes mellitus: a critical appraisal. J Diabetes Investig 2019;10:699–705
nofi. A.C., A.G., and J.I. are employees of Novo Diabetes Metab Syndr 2019;13:2126–2141 26. Banerji MA, Dunn JD. Impact of glycemic
Nordisk. A.C. and A.G. hold share options and/or 11. Polonsky WH, Fisher L, Hessler D, Bruhn D, control on healthcare resource utilization and
shares of Novo Nordisk. Novo Nordisk funded Best JH. Patient perspectives on once-weekly costs of type 2 diabetes: current and future
the trial and was responsible for trial design and medications for diabetes. Diabetes Obes Metab pharmacologic approaches to improving outcomes.
data analysis. No other potential conflicts of in- 2011;13:144–149 Am Health Drug Benefits 2013;6:382–392
terest relevant to this article were reported. 12. H€ ovelmann U, Brønsted L, Kristensen NR, 27. Boye KS, Curtis SE, Lage MJ, Garcia-Perez LE.
Author Contributions. All authors had full et al. Insulin icodec, an insulin analog suited for Associations between adherence and outcomes
access to all data, were responsible for data once-weekly dosing in type 2 diabetes (Abstract). among older, type 2 diabetes patients: evidence
interpretation and manuscript preparation, Diabetes 2020;69(Suppl. 1):237-OR from a Medicare Supplemental database. Patient
and had final responsibility for the decision to 13. Nishimura E, Kjeldsen T, Hubalek F, et al. Prefer Adherence 2016;10:1573–1581
submit for publication. H.S.B. is the guarantor Molecular and biological properties of insulin 28. Garcıa-Perez LE, Alvarez M, Dilla T,
of this work and, as such, had full access to icodec, a new insulin analog designed to give a Gil-Guillen V, Orozco-Beltran D. Adherence to
all of the data in the study and takes respon- long half-life suitable for once-weekly dosing therapies in patients with type 2 diabetes.
sibility for the integrity of the data and the (Abstract). Diabetes 2020;69(Suppl. 1):236-OR Diabetes Ther 2013;4:175–194
accuracy of the data analysis. 14. Rosenstock J, Bajaj HS, Janez A, et al.; 29. Khunti K, Millar-Jones D. Clinical inertia to
Prior Presentation. This study was presented NN1436-4383 Investigators. Once-weekly insulin insulin initiation and intensification in the UK: a
in e-poster form at the 56th Annual Meeting for type 2 diabetes without previous insulin focused literature review. Prim Care Diabetes
of the European Association for the Study of treatment. N Engl J Med 2020;383:2107–2116 2017;11:3–12
Diabetes, 21–25 September 2020. 15. American Diabetes Association. 6. Glycemic 30. Paul SK, Klein K, Thorsted BL, Wolden ML,
targets: Standards of Medical Care in Diabetes— Khunti K. Delay in treatment intensification
References 2019. Diabetes Care 2019;42(Suppl. 1):S61–S70 increases the risks of cardiovascular events in
1. American Diabetes Association. 9. Pharmacologic 16. Little RJA, Rubin DB. Statistical Analysis with patients with type 2 diabetes. Cardiovasc
approaches to glycemic treatment: Standards Missing Data. New York, John Wiley & Sons, 1987 Diabetol 2015;14:100