Transcript Autonomy Trial - Diabetes In Control

AUTONOMY: The First Randomized
Trial Comparing Two Patient-driven
Approaches to Initiate and Titrate
Prandial Insulin Lispro in
Type 2 Diabetes
Steve V. Edelman,1Rong Liu,2 Jennal Johnson,2 Leonard C. Glass2
Edelman et al. Diabetes Care 2014;37(8):2132-40.
This information may contain a substantive discussion of developmental
compound(s) not approved for marketing in the United States and/or products
approved for marketing in the United States but for uses, dosages, formulations
and/or populations different than those discussed in these materials. If the
information contains a substantive discussion of a currently marketed
product(s), a URL address to the U.S. Prescribing Information for that
product(s) is provided immediately below. You should consult this prescribing
information for the product’s approved uses and important information,
including boxed warnings, regarding the product’s use.
Humalog (insulin lispro injection, USP [rDNA origin]) Product Information
http://pi.lilly.com/us/humalog-pen-pi.pdf
AUTONOMY: Introduction (Slide 1 of 2)
 Management of patients with type 2 diabetes generally requires
stepwise intensification of therapy1,2:
• Lifestyle changes, OADs, and noninsulin injectable antidiabetic agents
• Given the progressive deterioration in β-cell function, progressing to
addition of exogenous insulin
 Results of the United Kingdom Prospective Diabetes Study
(UKPDS) support the need for treatment intensification with
exogenous insulin in combination with OADs in a significant
percentage of patients to achieve and maintain metabolic
control3
1. Inzucchi et al. Diabetes Care 2012;35(6):1364-79(Updated 36: p 490).
2. Garber et al. Endocr Pract 2013;19(2):327-36.
3. Turner et al. JAMA 1999;281(21):2005-12.
AUTONOMY: Introduction (Slide 2 of 2)
 Given limited data and myriad treatment approaches, there is
currently no global clinical consensus for the approach to
treatment intensification with insulin therapy
 Meta-analysis indicates most effective use of insulin is achieved
using a basal/bolus regimen1
 Currently, there is great interest in sequential addition of prandial
insulin in patients with type 2 diabetes not attaining adequate
glycemic control with basal insulin only2,3
1. Giugliano et al. Diabetes Care 2011;34(2):510-7.
2. Rodbard et al. Lancet Diabetes Endocrinol 2014;2(1):30-7.
3. Edelman et al. Diabetes Care 2014;37(8):2132-40.
AUTONOMY: Study Objective
 To compare efficacy and safety of 2 patient-based
self-titration algorithms for initiation and titration of prandial insulin
lispro in patients with type 2 diabetes inadequately controlled on
basal insulin plus OADs in endocrine and generalist settings
• Approximately 44% of trial sites were in primary care settings
 1st comparison of 2 self-titration insulin algorithms for escalation of
prandial insulin lispro therapy in a large, multicountry, randomized,
controlled trial
• Designed as 2 independent parallel studies (Study A, N = 528; Study B,
N = 578) utilizing a single protocol to corroborate substantial evidence
of reproducibility; data for each study were analyzed separately and
independently
Edelman et al. Diabetes Care 2014;37(8):2132-40.
AUTONOMY: Study Design
NO
Enrollment
Optional: Insulin
Glargine (GLA)
Optimization
Lead-ina
Add insulin lispro 1-2-3
with adjustments
(Q1D)
Randomize
HbA1c ≤7.0%
(≤53.0 mmol/mol)
Add insulin lispro 1-2-3
with adjustments
(Q3D)
YES
Discontinue
6 Weeks
Visit:
24 Weeks
1 (screening) 2a 3a 4a,b 5a,b 6a
Week: -1
a6-week
0
1
2
4
6
7a
8b
9b
10
11b 12
13b 14b 15
16b 17
18b 19
7
8
9
10
12
16
23
29
14
18
19
27
31
GLA optimization lead-in only required for subjects who had to be converted to GLA from insulin NPH, ILPS,
or detemir; required conversion from GLA twice daily to once daily; or those on once-daily GLA at study entry with
HbA1c >7.0% (>53.0 mmol/mol) and fasting blood glucose >120 mg/dL (>6.7 mmol/L). Subjects who did not require GLA
optimization were randomized at Visit 2, forewent Visits 3 to 7 and instead proceeded to the randomized treatment period
beginning with Visit 8 activities, 1 week after Visit 2; bTelephone visits
Edelman et al. Diabetes Care 2014;37(8):2132-40.
AUTONOMY: Treatment Algorithms
Randomize 1:1
Q1D
Q3D
Self-titrated daily based on premeal glucose
readings from prior day (eg, prebreakfast dose
adjusted using prior day prelunch reading)
Self-titrated every 3 days based on median
blood glucose readings from the 3 days prior
(eg, prebreakfast dose adjusted using median
prelunch blood glucose reading from the past 3 days)
Premeal Target Blood Glucose: 85-114 mg/dL
Blood Glucose
Median Reading:
Blood Glucose
Daily Reading:
85-114 mg/dL
No change in dose
85-114 mg/dL
No change in dose
Target not achieved
Increase dose by 1 U/day
until target is reached
115-144 mg/dL
Dose increased by 2U
≥145 mg/dL
Dose increased by 4U
56-84 mg/dL
Dose decreased by 1U
56-84 mg/dL
Dose decreased by 2U
<56 mg/dL
Dose decreased by 2U
<56 mg/dL
Dose decreased by 4U
Starting insulin lispro dose: 10% of total insulin glargine dose; first dose administered with breakfast unless subject does
not eat breakfast, then first dose administered at lunch time at investigator discretion
Edelman et al. Diabetes Care 2014;37(8):2132-40.
AUTONOMY: Inclusion Criteria
 Type 2 diabetes
 18-85 years of age
 BMI <45 kg/m2
 HbA1c >7.0% and ≤12.0% (>53.0 and ≤107.7 mmol/mol)
 Treated with ≥20 U/d of insulin glargine (GLA), neutral
protamine Hagedorn (NPH), insulin lispro protamine suspension
(ILPS), or detemir; and had been using metformin, meglitinide,
sulfonylurea, pioglitazone, sitagliptin, or a combination of these
for ≥3 months
• Meglitinides or sulfonylureas were discontinued at randomization
Edelman et al. Diabetes Care 2014;37(8):2132-40.
AUTONOMY: Exclusion Criteria







Prior rapid- or short-acting insulin therapy
Excessive insulin resistance (>2 U/kg/day)
Morbid obesity (BMI ≥45 kg/m2)
Pregnancy or planned pregnancy
Cancer
Recent history of severe hypoglycemia
Moderate to severe cardiovascular, renal, hepatic, or
hematologic disease
 Treatment with GLP-1 receptor agonists, α-glucosidase
inhibitors, DPP-4 inhibitors (except sitagliptin), and
rosiglitazone within 3 months, or glucocorticoids within
2 weeks of screening
Edelman et al. Diabetes Care 2014;37(8):2132-40.
AUTONOMY: Primary and
Secondary Outcome Measures
 Primary efficacy outcome:
•
Compare change in HbA1c from baseline to endpoint (Week 24 after
randomization) for Q1D and Q3D algorithms
 Secondary outcomes:
•
Incidence and annualized rate of self-reported total, severe, and nocturnal
hypoglycemia
•
Proportion of subjects achieving target values HbA1c ≤7.0%
•
Change in FBG, 7-point SMBG profile, and weight from baseline
•
Change in dose of basal (GLA) and prandial (lispro) insulin at end of study
•
Change in 1,5-anhydroglucitol (a marker of hyperglycemia)
 Assessments in subjects ≥65 years of age:
•
Change in HbA1c, hypoglycemia (incidence and rate), FBG, and proportion of
subjects achieving target
 Safety was monitored throughout the study (hypoglycemia was
considered an AE with severe hypoglycemia recorded as an SAE)
Edelman et al. Diabetes Care 2014;37(8):2132-40.
AUTONOMY: Statistical Analyses
(Slide 1 of 3)
 A classification method was applied for primary efficacy analysis, as no prior
projection, preference, or historical evidence regarding which self-titration
algorithm performs better was available1
 All safety outcomes were assessed in the entire randomized population (ie,
all subjects who entered the study, completed the GLA optimization lead-in
period [if applicable], and were randomized)
 All efficacy analyses were based upon the full analysis set (subjects in the all
randomized population who took at least 1 dose of insulin lispro), and a
sensitivity analysis was conducted for the primary efficacy measure based
upon the all-completer population
 All efficacy and safety analyses were conducted at an α-level of .05
 All CIs were computed as 2-tailed using a 95% significance level
 For categorical measures, including AEs and hypoglycemia incidence,
Fisher’s exact test or Pearson’s chi-square test was used
1. Qu et al. Stat Med 2011;30(30):3488-95.
Edelman et al. Diabetes Care 2014;37(8):2132-40.
AUTONOMY: Statistical Analyses
(Slide 2 of 3)
 Continuous efficacy and safety variables measured repeatedly were
evaluated using a mixed model repeated measure (MMRM)
approach using the restricted maximum likelihood method, including
the following independent variables:
• Fixed effects for treatment algorithm, all stratification variables, visit,
treatment-by-visit interaction, and baseline outcome variable as the
covariate
 Treatment-by-age group (≥65 years, <65 years) interaction for the
change in HbA1c was tested using another MMRM model with
additional items, including:
• Subgroup and subgroup-by-treatment algorithm interaction
Edelman et al. Diabetes Care 2014;37(8):2132-40.
AUTONOMY: Statistical Analyses
(Slide 3 of 3)
 Hypoglycemia incidence was analyzed with a logistic model with terms for
treatment algorithm and all stratification variables as sensitivity analysis
 The rate of total, nocturnal, and severe hypoglycemia per year during the
treatment phase was analyzed using LOCF, applying a negative binomial
model with terms for treatment algorithm, HbA1c stratum,
sulfonylurea/meglitinide use, and the logarithm of the exposure time (in
days) as an offset variable and compound symmetry as variancecovariance structure
 A Wilcoxon rank-sum test was conducted as a sensitivity analysis
 The percentages of subjects achieving HbA1c targets at the end of the
study (LOCF) were analyzed using a logistic regression model with terms
for treatment algorithm and strata
 All data are expressed as least square mean ± standard error unless
otherwise stated, and a p-value of <.05 was considered significant
Edelman et al. Diabetes Care 2014;37(8):2132-40.
AUTONOMY: Patient Disposition of
All Randomized Subjects
Study B
Randomized (N = 581)
Study A
Randomized (N = 531)
Q1D (N = 268)a
Q3D (N = 263)a
Q1D (N = 289)a
Completed, n = 223 (83.2%)
Completed, n = 210 (79.8%)
Early terminated, n = 45 (16.8%)
Early terminated, n = 53 (20.2%)
Completed, n = 244 (84.4%)
Early terminated, n = 45 (15.6%)
Q3D (N = 292)a
Completed, n = 241 (82.5%)
Early terminated, n = 51 (17.5%)
Reasons:
n
%
Reasons:
n
%
Reasons:
n
%
Reasons:
n
%
Adverse event
1
0.4
Adverse event
4
1.5
Adverse event
2
0.7
Adverse event
3
1.0
Death
2
0.7
Death
0
0.0
Death
1
0.3
Death
3
1.0
Entry criteria not met
1
0.4
Entry criteria not met
3
1.1
Entry criteria not met
4
1.4
Entry criteria not met
7
2.4
Lack of efficacy
1
0.4
Lack of efficacy
2
0.8
Lack of efficacy
1
0.3
Lack of efficacy
0
0.0
Lost to follow-up
4
1.5
Lost to follow-up
4
1.5
Lost to follow-up
8
2.8
Lost to follow-up
9
3.1
Physician decision
4
1.5
Physician decision
10
3.8
Physician decision
8
2.8
Physician decision
11
3.8
13
4.9
Protocol violation
8
2.8
Protocol violation
8
2.7
Protocol violation
17
6.3
Protocol violation
Sponsor decision
1
0.4
Sponsor decision
2
0.8
Sponsor decision
0
0.0
Sponsor decision
1
0.3
Subject decision
14
5.2
Subject decision
15
5.7
Subject decision
13
4.5
Subject decision
9
3.1
a6
subjects were randomized but did not receive at least 1 dose of lispro: Study A Q1D (n = 1) and Q3D (n = 2);
Study B Q1D (n = 1) and Q3D (n = 2). These subjects (ie, subjects not exposed to lispro) were not included in the full
analysis set. Patient disposition was based on all randomized subjects. There was no significant difference between
the percentages of subjects who discontinued from Q1D or Q3D for any reason of early termination. Deaths were not
attributed to the treatment
Edelman et al. Diabetes Care 2014;37(8):2132-40.
AUTONOMY: Baseline Demographics
Baseline Demographics
Age, years, mean (SD)
Subjects ≥65 years, %
Caucasian, %
Female, %
BMI, kg/m2, mean (SD)
BMI ≥30kg/m2, %
Body weight, kg (SD)
Duration of diabetes,
years, mean (SD)
Duration >10 years
HbA1c %, mean (SD)
Study A
Study B
Q1D
Q3D
Q3D vs. Q1D
Q1D
Q3D
Q3D vs. Q1D
(N = 267) (N = 261)
p-value
(N = 288) (N = 290)
p-value
57.9 (10.3) 58.8 (9.5)
.278
57.7 (9.7) 57.0 (10.6)
.412
24.3
26.4
.618
19.4
22.4
.414
82.3
83.5
.781
79.7
83.3
.308
49.8
52.9
.487
53.8
53.4
.934
33.3 (5.3) 33.4 (5.5)
.793
32.6 (5.2) 33.2 (5.7)
.174
73.4
69.7
.385
66.3
68.6
.594
94.6 (20.2) 92.4 (17.7)
.188
90.8 (18.3) 93.5 (21.2)
.112
11.7 (6.3)
12.6 (7.9)
.129
11.6 (6.5)
11.9 (7.1)
.645
54.3
8.3 (0.9)
60.2
8.4 (1.0)
.188
.453
54.5
8.3 (1.0)
53.8
8.4 (1.0)
.868
.162
mmol/mol, mean (SD)
67.2 (9.8)
68.3 (10.9)
-
HbA1c >8.0%
(>63.93 mmol/mol), %
56.6
58.2
.725
67.2 (10.9) 68.3 (10.9)
53.5
57.6
.357
p-values for continuous measures were based on an analysis of variance, and categorical measures
were based on Fisher’s exact test for treatment algorithm Q3D vs. Q1D
Edelman et al. Diabetes Care 2014;37(8):2132-40.
AUTONOMY: Concomitant Medications
Study A
Study B
Q1D
(N = 267)
Q3D
(N = 261)
Q1D
(N = 288)
Q3D
(N = 290)
Biguanides, %
85.4
89.3
93.8
89.3
Sulfonylurea/meglitinide, %
49.4
52.5
34.7
40.3
DPP-4 inhibitors, %
9.7
10.0
8.0
7.2
Thiazolidinediones, %
5.2
7.3
3.8
6.6
OAD class ≥2, %
44.9
51.0
36.1
39.3
Concomitant Medications
p-values were not calculated
Edelman et al. Diabetes Care 2014;37(8):2132-40.
AUTONOMY: Change from Baseline
in HbA1c
Study A: HbA1c (%)
Study B: HbA1c (%)
0.0
Q1D
Q3D
N = 528
-0.2
-0.4
-0.6
*
-0.8
*
-1.0
*
-1.2
Baseline
Week 7
Week 12 Week 24
Q3D-Q1D LSM:
0.07
0.08
95% CI: -0.04, 0.17 -0.06, 0.22
0.04
-0.15, 0.22
HbA1c Change from Baseline
(%, LSM SE)
HbA1c Change from Baseline
(%, LSM SE)
0.0
Q1D
Q3D
N = 578
-0.2
-0.4
-0.6
*
-0.8
*
-1.0
*
-1.2
Baseline
Week 7
Week 12 Week 24
Q3D-Q1D LSM:
-0.01
-0.02
95% CI: -0.11, 0.09 -0.15, 0.11
0.06
-0.12, 0.24
*Significant change from baseline based on 95% CIs from an MMRM approach using restricted maximum likelihood
method (REML) for both Q1D and Q3D
Edelman et al. Diabetes Care 2014;37(8):2132-40.
AUTONOMY: Overall Percentage of
Subjects with HbA1c ≤7.0% at 24 Weeks
Study A (N = 528)
Percentage of Subjects with
HbA1c ≤7.0% at 24 Weeks
90%
100%
p = .128
90%
80%
70%
60%
50%
49.8%
42.5%
40%
30%
20%
Percentage of Subjects with
HbA1c ≤7.0% at 24 Weeks
100%
Study B (N = 578)
80%
70%
60%
50%
42.4%
30%
20%
10%
0%
0%
Q3D
49.3%
40%
10%
Q1D
p = .162
Q1D
Q3D
No statistically significant difference between Q1D and Q3D algorithms for Study A and Study B for
overall percentage of subjects reaching target HbA1c ≤7.0% (≤53.0 mmol/mol) at end of study (24 weeks)
Edelman et al. Diabetes Care 2014;37(8):2132-40.
AUTONOMY: Subjects ≥65 Years of Age
with HbA1c ≤7.0% at 24 Weeks
Study A (N = 134)1
Study B (N = 121)1
100%
p = .701
90%
80%
70%
60%
58.5%
58.0%
50%
40%
30%
20%
Percentage of Subjects with
HbA1c ≤7.0% at 24 Weeks
Percentage of Subjects with
HbA1c ≤7.0% at 24 Weeks
90%
100%
*p = .015
80%
70%
67.9%
60%
*
50%
46.2%
40%
30%
20%
10%
10%
0%
0%
Q1D
Q3D
Q1D
Q3D
(n = 65)1
(n = 69)1
(n = 56)1
(n = 65)1
In Study B, percentage of subjects ≥65 years of age reaching target HbA1c ≤7.0% (≤53.0 mmol/mol) at end of study
(24 weeks) was significantly lower for those randomized to Q3D (46.2%) than to Q1D (67.9%), p = .015; in Study A, there
was no statistical difference between Q1 vs. Q3 algorithms
Edelman et al. Diabetes Care 2014;37(8):2132-40.
1. Data on file, Eli Lilly and Company.
AUTONOMY: SMBG Profiles at Baseline
and Week 24
300
Study B (N = 578)
Q1D Baseline
Q1D Week 24
Q3D Baseline
Q3D Week 24
250
200
150
100
50
0
*
*
*
*
*
*
7-Point Self-monitored Blood Glucose (mg/dL)
Mean SD
7-Point Self-monitored Blood Glucose (mg/dL)
Mean SD
Study A (N = 528)
300
250
Q1D Baseline
Q1D Week 24
Q3D Baseline
Q3D Week 24
200
150
100
50
*
*
†
*
*
*
*
†
†
0
*Significant change in SMBG from baseline based on 95% CIs from an MMRM approach using restricted maximum likelihood method
(REML) for both Q1D and Q3D; †Significant difference between Q3D and Q1D at Week 24 from an MMRM model using REML
Edelman et al. Diabetes Care 2014;37(8):2132-40.
AUTONOMY: Change from Baseline
in 1,5-AG
Study A: 1,5-AG (μg/mL)
Study B: 1,5-AG (μg/mL)
3.0
4.0
Q1D
Q3D
N = 528
*
2.0
*
1.0
*
0.0
1,5-AG Change from Baseline
(μg/mL, LSM SE)
1,5-AG Change from Baseline
(μg/mL, LSM SE)
4.0
3.0
Q1D
Q3D
N = 578
*
2.0
*
1.0
*
0.0
Baseline
Week 7
Week 12 Week 24
Q3D-Q1D LSM:
-0.08
-0.29
95% CI: -0.54, 0.38 -0.88, 0.30
-0.15
-0.95, 0.66
Baseline
Week 7
Week 12 Week 24
Q3D-Q1D LSM:
0.10
0.09
95% CI: -0.33, 0.53 -0.51, 0.70
-0.28
-1.08, 0.52
*Significant change from baseline based on 95% CIs from an MMRM approach using restricted maximum likelihood
method (REML) for both Q1D and Q3D
Edelman et al. Diabetes Care 2014;37(8):2132-40.
AUTONOMY: Overall Hypoglycemia
Incidences and Rates per 1 Year
Hypoglycemia
Total:
Q1D
(N = 268)
Study A
Q3D vs. Q3D/Q1D
Q3D
Q1D
Rate Ratio
(N = 263)
p-value
(95% CI)
Q1D
(N = 289)
Study B
Q3D vs. Q3D/Q1D
Q3D
Q1D Rate Ratio
(N = 292) p-value (95% CI)
Incidence, n (%)
231
(86.2)
218
(83.2)
.435
-
238
(82.4)
231
(79.1)
.351
-
Rate per 1 year,NBM (SE)
38.32
(2.80)
40.58
(3.06)
.586
1.06
(0.86-1.30)
38.76
(3.14)
40.54
(3.29)
.689
1.05
(0.84-1.30)
Incidence, n (%)
194
(72.4)
191
(72.9)
.923a
.842b
-
205
(70.9)
185
(63.4)
.053a
.032b
-
Rate per 1 year,NBM (SE)
20.88
(1.93)
24.16
(2.30)
.272c
.504d
1.16
(0.89-1.50)
18.22
(1.79)
21.26
(2.11)
.260c
.561d
1.17
(0.89-1.53)
Incidence, n (%)
169
(63.1)
167
(63.7)
.870
-
156
(54.0)
149
(51.0)
.470
-
Rate per 1 year,NBM (SE)
8.59
(0.80)
9.60
(0.93)
.404
1.12
(0.86-1.45)
7.14
(0.80)
8.23
(0.91)
.358
1.15
(0.85-1.56)
Documented symptomatic1:
Nocturnal:
Incidence is reported as the number of subjects with at least 1 hypoglycemic episode; p-values for the incidences of each category
were based on a logistic regression model for Q3D vs. Q1D; p-values for rate adjusted per 1 year were based on NBM regression for
Q3D vs. Q1D; documented symptomatic p-values calculated by aFisher exact test, bLogistic regression model, cNegative binomial
regression model, and dWilcoxon rank sum test
1. Data on file, Eli Lilly and Company.
Edelman et al. Diabetes Care 2014;37(8):2132-40.
AUTONOMY: Overall Severe Hypoglycemia
Incidences and Rates per 1 Year
Study A
Q3D vs.
Q1D
p-value
Study B
Q3D/Q1D
Rate Ratio
Q1D
(95% CI) (N = 289)
Q3D/Q1D
Rate Ratio
(95% CI)
Q1D
(N = 268)
Q3D
(N = 263)
Incidence, n (%)
5
(1.9)
2
(0.8)
.258
-
7
(2.4)
8
(2.7)
.856
-
Rate per 1 year, mean (SD)
0.04
(0.31)
0.03
(0.41)
.271
-
0.11
(1.09)
0.06
(0.36)
.816
-
Hypoglycemia
Severe:
Q3D
(N = 292)
Q3D vs.
Q1D
p-value
Incidence is reported as the number of subjects with at least 1 hypoglycemic episode; p-values for incidences were
based on a logistic regression model for Q3D vs. Q1D; due to low occurrence of severe hypoglycemia, mean ± SD
and only Wilcoxon test p-values are presented
Edelman et al. Diabetes Care 2014;37(8):2132-40.
AUTONOMY: Hypoglycemia Incidences and
Rates per 1 Year for Subjects ≥65 Years of Age
Hypoglycemia
Total:
Q1D
(N = 66)
Study A
Q3D vs. Q3D/Q1D
Q3D
Q1D
Rate Ratio
(N = 69)
p-value
(95% CI)
Q1D
(N = 56)
Study B
Q3D vs. Q3D/Q1D
Q3D
Q1D Rate Ratio
(N = 65)
p-value (95% CI)
Incidence, n (%)
60
(90.9)
61
(88.4)
.802
-
51
(91.1)
53
(81.5)
.205
-
Rate per 1 year,NBM (SE)
41.62
(5.42)
48.84 (6.21)
.383
1.17
(0.82-1.68)
51.38
(8.26)
42.88
(6.35)
.404
0.83
(0.55-1.28)
Incidence, n (%)
52
(78.8)
56
(81.2)
.831a
.617b
-
39
(69.6)
41
(63.1)
.564a
.265b
-
Rate per 1 year,NBM (SE)
20.84
(3.51)
28.30
(4.63)
.197c
.546d
1.36
(0.85-2.16)
22.48
(5.13)
19.06
(4.05)
.593c
.373d
0.85
(0.46-1.55)
Incidence, n (%)
45
(68.2)
49
(71.0)
.763
-
42
(75.0)
43
(66.2)
.383
-
Rate per 1 year,NBM (SE)
8.71
(1.48)
11.60 (1.92)
.229
1.33
(0.84-2.12)
12.01
(2.40)
10.69 (2.03)
.671
0.89
(0.52-1.52)
Documented symptomatic1:
Nocturnal:
Incidence is reported as the number of subjects with at least 1 hypoglycemic episode; p-values for the incidences of each category
were based on a logistic regression model for Q3D vs. Q1D; p-values for rate adjusted per 1 year were based on NBM regression for
Q3D vs. Q1D; documented symptomatic p-values calculated by aFisher exact test, bLogistic regression model, cNegative binomial
regression model, and dWilcoxon rank sum test
1. Data on file, Eli Lilly and Company.
Edelman et al. Diabetes Care 2014;37(8):2132-40.
AUTONOMY: Severe Hypoglycemia Incidences and
Rates per 1 Year for Subjects ≥65 Years of Age
Study A
Q3D vs.
Q1D
p-value
Q3D/Q1D
Rate Ratio
(95% CI)
Q1D
(N = 66)
Q3D
(N = 69)
Incidence, n (%)
3
(4.5)
1
(1.4)
.296
Rate per 1 year, mean (SD)
0.10
(0.49)
0.03
(0.25)
.294
Hypoglycemia
Severe:
Study B
Q3D vs.
Q1D
p-value
Q3D/Q1D
Rate Ratio
(95% CI)
Q1D
(N = 56)
Q3D
(N = 65)
-
1
(1.8)
2
(3.1)
.797
-
-
0.05
(0.37)
0.07
(0.38)
.657
-
Incidence is reported as the number of subjects with at least 1 hypoglycemic episode; p-values for incidences were
based on a logistic regression model for Q3D vs. Q1D; due to low occurrence of severe hypoglycemia, mean ± SD
and only Wilcoxon test p-values are presented
Edelman et al. Diabetes Care 2014;37(8):2132-40.
*p = .014
*
Weight (kg) Change from Baseline
(LSM SE) at 24 Weeks
Weight (kg) Change from Baseline
(LSM SE) at 24 Weeks
AUTONOMY: Weight Change from
Baseline at 24 Weeks
p = .108
In both studies, subjects gained weight from baseline to endpoint (24 weeks) regardless of titration algorithm;
In Study A, subjects using Q3D algorithm gained more weight from baseline than subjects using Q1D algorithm
(p = .014); no significant difference in weight gain was observed between Q3D and Q1D in Study B
Edelman et al. Diabetes Care 2014;37(8):2132-40.
AUTONOMY: Insulin Dose
Study A
Insulin Dose (U/d)
GLA at entry, n
mean (SD)
Q1D
(N = 267)
180
46.8 (32.4)
Q3D
(N = 261)
177
48.6 (27.8)
Study B
Q3D vs.
Q1D
p-value
-
Q1D
(N = 288)
163
46.8 (29.2)
Q3D
(N = 290)
163
45.0 (30.0)
Q3D vs.
Q1D
p-value
-
NPH at entry, n
50
48
75
78
mean (SD)
50.4 (26.7) 45.0 (26.4)
46.1 (32.7) 47.6 (23.2)
Detemir at entry, n
37
35
49
47
mean (SD)
58.9 (42.6) 46.2 (30.8)
52.4 (32.0) 60.8 (44.1)
ILPS at entry, n
0
0
0
1
mean (SD)
NA
NA
NA
34.0 (NA)
Basal (GLA)
atrandomization, mean
62.8 (33.9) 60.3 (32.1)
.335
57.3 (32.5) 60.0 (33.0)
.236
(SD)
Basal (GLA) at Week 24,
66.4 (35.1) 63.5 (34.6)
.543
59.9 (33.4) 65.2 (42.5)
.497
mean (SD)
Bolus (lispro) at Week 24,
47.7 (41.1) 54.6 (46.7)
.095
44.5 (36.8) 48.8 (51.0)
.156
mean (SD)
p-values for continuous measures were based on an analysis of variance, and categorical measures
were based on Fisher’s exact test for treatment algorithm Q3D vs. Q1D
Edelman et al. Diabetes Care 2014;37(8):2132-40.
AUTONOMY: Bolus Injections at 24-Week
Endpoint
Study A
Study B
Bolus Injections
(LOCF Subjects)
Q1D
(N = 267)
Q3D
(N = 261)
Q1D
(N = 288)
Q3D
(N = 290)
1 injection, n (%)
84 (31.5)
81 (31.0)
102 (35.4)
100 (34.5)
2 injections, n (%)
69 (25.8)
66 (25.3)
85 (29.5)
89 (30.7)
3 injections, n (%)
114 (42.7)
114 (43.7)
101 (35.1)
101 (34.8)
p-values were not calculated
Edelman et al. Diabetes Care 2014;37(8):2132-40.
AUTONOMY: Discussion (Slide 1 of 2)
 Both algorithms (Q1D, Q3D) demonstrated statistically significant
and clinically equivalent reductions in HbA1c, significant increases
in 1,5-AG, and improved 7-point SMBG profiles in Studies A and B
 ~50% of subjects who had previously failed to reach goal HbA1c of
≤7.0% (53.0 mmol/mol) with basal insulin optimization plus OADs
achieved the ADA goals for glycemic control with less glucose
variability
 Sequential addition of prandial insulin lispro injections resulted in
~61% of subjects only requiring ≤2 doses rather than a full basalbolus regimen (ie, simplifies treatment, could enhance therapy
compliance)
 Subjects gained 2-3 kg of weight, regardless of treatment algorithm,
with the initiation of prandial insulin
Edelman et al. Diabetes Care 2014;37(8):2132-40.
AUTONOMY: Discussion (Slide 2 of 2)
 Results show basal-bolus therapy can be initiated in the elderly
without increased risk of hypoglycemia
 Regardless of titration algorithm, improved metabolic control was
accomplished with low incidences and rates of nocturnal and severe
hypoglycemia in both the overall study population and the elderly
subgroup (≥65 years of age) with initiation and escalation of lispro
Edelman et al. Diabetes Care 2014;37(8):2132-40.
AUTONOMY: Limitations
 Exclusion of subjects with BMI ≥45 kg/m2, a potentially important
population with the growing healthcare burden associated with
obesity
 Safety and efficacy need to be addressed in future research for
Asian populations, as no Asian countries were included
 Although numerical differences were observed, which seem to
benefit Q1D vs. Q3D, these were not statistically significant and
should only be considered hypothesis-generating
 AUTONOMY algorithms were based on PK/PD modeling of GLA
and lispro insulins; there may not be substantial differences with the
use of other short-acting prandial insulin analogues
 Other combinations to control postprandial glucose excursions may
be considered (such as combination of GLP-1 agonists with insulin)
Edelman et al. Diabetes Care 2014;37(8):2132-40.
AUTONOMY: Conclusions
 Trial provided novel data and basis for initiation and escalation of
lispro therapy using 2 simple, self-titration regimens in patients with
type 2 diabetes who failed to achieve adequate glycemic control on
appropriately titrated basal insulin plus OADs
 Trial demonstrated a basal-bolus regimen can be initiated in an
adult population, including the elderly, to lower HbA1c and limit
mealtime glucose excursions safely, with either patient-driven
algorithm, in endocrinology and generalist settings
 Utilizing Q1D and Q3D algorithms simplified insulin therapy by not
requiring patient training on carbohydrate counting or insulin
correction factor, and reduced the number of OADs in those treated
with sulfonylurea or meglitinide
Edelman et al. Diabetes Care 2014;37(8):2132-40.