Presented at the Keck School of Medicine of USC Research Seminar in October 2011. Learn more about Tom's latest work: http://profiles.sc-ctsi.org/thomas.buchanan
Disentangling the origin of chemical differences using GHOST
Does making too much insulin cause type 2 diabetes?
1. Does Making too much Insulin
Cause Type 2 Diabetes?
Thomas A. Buchanan, MD
For the
USC Gestational Diabetes Study Group
2. Teach you something about the biology,
prediction and prevention of type 2 diabetes
Show you what an interdisciplinary team
working in humans can accomplish
Goals
3. The USC Gestational Diabetes Study Group
Research Staff
Enrique Trigo
Karla Garcia
Lilit Baronikian
Miwa Kawakubo
Aura Marroquin
Cesar Ochoa
Jose Goico
Sylvia Tan
Investigators
Tom Buchanan
Siri Kjos
Richard Watanabe
Stan Azen
Anny Xiang
Ruth Peters
Hooman Allayee
Jean Lawrence
Katie Page Penina Segall-Gutierrez
Jorge Caro George Aurea
Isabel Enriquez
Participants: GDM Cohort Study, TRIPOD and PIPOD Studies, BetaGene Study
With Strong Support from:
Funding: NIDDK, NCRR/GCRC/CTSI, ADA, Parke-Davis,
Takeda
Collaborators: Bergman group, Howard Hodis, Wendy Mack, Goran group
5. Definition
Glucose intolerance with onset or first recognition
during pregnancy
Gestational Diabetes Mellitus
Detection
Screen pregnant women not known to have diabetes:
for at-risk clinical characteristics
for diagnostic oral glucose tolerance
“Population” screening for elevated glucose levels in young womenGlucose levels that might cause fetal morbidity
7. 0
10
20
30
40
50
60
70
80
5 10 15 20 25 30
Navajo
o
o
o
o
o
o
o Zuni
Mixed/Other
Hispanic (USC)
x
x
x
x
x
x x
x
x
x Boston Cohort
Years after Delivery
CumulativeIncidenceofDiabetes(%)
Diabetes After GDM
Kim et al: Diabetes Care, 2002
8. Multifaceted Approach to Diabetes after GDM
USC Gestational Diabetes Study Group
Clinical Cohort
Women followed in standard clinical care with annual testing for diabetes after
index pregnancy
Main Outcomes: diabetes incidence and clinical risk factors
Physiological Cohort
Women followed with detailed physiological testing during pregnancy and at 15-
month intervals thereafter
Main Outcomes: physiological mechanisms for development of diabetes
Interventional Cohort
Women enrolled in clinical trials of diabetes prevention and early treatment
Main Outcomes: feasibility and mechanisms for diabetes prevention
Genetic Cohort
GDM and control probands and their families studied with detailed physiological
and morphological phenotyping and genetic analysis
Main Outcomes: genetic associations with diabetes quantitative traits
9. Defining Diabetes Incidence and Clinical Risk Factors
Subjects: Hispanic women with GDM who delivered at
Womens Hospital (n=700-1000)
Setting: Outpatient clinic where women returned for diabetes
testing postpartum and annually
Design: Observational
Main Outcomes: diabetes incidence rate and clinical risk
factors for diabetes
Investigators: Siri Kjos, Anny Xiang, Ruth Peters, Tom
Buchanan
Support: faculty blood, sweat and tears!
Clinical Cohort
10. Kjos et al: Diabetes 44:586-591, 1994
Diabetes after GDM
Cumulative Incidence in Clinical Cohort
Years Postpartum
CumulativeIncidence(%)
0
20
40
80
60
0 1 2 3 4 5 6
n=671
11. Diabetes after GDM
Risk Factors in Clinical Cohort
At Baseline (pregnancy and immediate postpartum)
Early gestational age at diagnosis1
High glucose levels1
During Follow-up
Weight gain2
Additional pregnancy2
Progestin-only contraception3,4
1
Kjos et al: Diabetes 44:586-591, 1995, 2
Peters et al: Lancet 347:227-230, 1996
3
Kjos et al: JAMA 280:533-538, 1998, 4
Xiang et al: Diabetes Care 29:613-617, 2006
Insulin Resistance
}
Hypothesis: Insulin resistance is causing β-cell failure
12. Identifying Physiological Determinants of Diabetes
Subjects: Prospectively recruited cohort of Hispanic women
with GDM who delivered at Womens Hospital (n=150)
Setting: GCRC-based study
Design: Observational - detailed physiological measurements
of glucose regulation during pregnancy and at 15-month
intervals thereafter
Main Outcomes: physiological changes that predict or attend
the development of diabetes
Investigators: Tom Buchanan, Anny Xiang, Ruth Peters, Siri
Kjos
Support: NIH R01 DK46374 (1993-2007)
Physiological Cohort
13. USC GDM Cohort Study
Overview of Study Design
Hispanic American women
GDM by 3rd
GDM Workshop
criteriaIslet cell antibody negative
Detailed Metabolic Measurements
Glucose levels
Insulin resistance
β-cell function
Body composition
15 30 45 60 75 900
Months After Delivery
3rd
TM
Non-Pregnant
105 120 132 144
n=150 GDM
n=30 Control
14. Gestational Diabetes Mellitus
Multiple Metabolic Defects during the Third Trimester
Xiang et al: Diabetes 48:848-854, 1999
Beta Cell Compensation
1500
1000
500
0
GDMControl
DispositionIndex
p<0.0001
IVGTT
Skeletal Muscle
Control (30)
GDM (150)
Adipose Tissue
Control
GDM
Liver
Control
GDM
15. Type 2 Diabetes after GDM
Cumulative Incidence in Physiological Cohort
Xiang et al: Diabetes 59:2652-2630, 2010
n=72 with
multiple visits
16. Resistant Sensitive
Regulation of Blood Glucose
Normal
Diabetic
Impaired
Insulin Sensitivity
InsulinSecretion Insulin Sensitivity and Secretion
Bergman et al: J Clin Invest, 1981
Normal
Abnormal
Sensitivity x Output = Constant
“Disposition Index”
Disposition Index
2000
1000
200
17. 0 1 2 3
0
200
400
600
800
Insulin Sensitivity (SI)
AcuteInsulinResponse
Yes (n=24) No (n=47)Diabetes:
Xiang et al: Diabetes 55:1074-1079, 2006
Longitudinal Changes: First Five Years
β-cell Function after GDM
3.9 years
3.7 years
18. Declining β-cell Function after GDM
Relation to Glucose Levels
Prior GDMs (n=71): OGTTs and IVGTTs at 15, 30, 45, 60, 75 months postpartum
Yes (n=24)No (n=47)Diabetes:
Diabetes
0 200 400 600 800 1000
0
100
200
Disposition Index
mg/dl
Fasting Glucose
0 200 400 600 800 1000
0
100
200
300
Disposition Index
mg/dl
Diabetes
Yes (n=24)No (n=47)Diabetes:
OGTT 2hr Glucose
Xiang et al: Diabetes 55:1074-1079, 2006 5-11-05
20. Evolution of Hyperglycemia after GDM
What predicts diabetes?
5-11-05
0 200 400 600 800 1000
0
100
200
300
Disposition Index
mg/dl
Diabetes
Yes (n=24)No (n=47)Diabetes:
OGTT 2hr Glucose
Answer: characteristics of people who almost have diabetes
(PLUS: falling β-cell function, weight gain, pregnancy, progestins)
22. Subjects: n=60 with at least two visits by 75 months
Baseline Variables: body mass and fat, glucose levels,
insulin levels, insulin resistance, β-cell compensation, lipid
levels (FFA and clinical lipids), adipocytokines
During Follow-up: pregnancy; hormonal contraception;
weight and fat gain; change in insulin sensitivity, lipids,
adipocytokines
Analysis: Random coefficients mixed modeling to identify
factors predictive of (for baseline variables) or associated
with (for follow-up variables) change β-cell compensation
Predicting Falling β-cell Function
Design
23. Only Independent Correlate: Weight Gain (p=0.003)
Results
Predicting Falling β-cell Function
Xiang et al: Diabetes Care 33:396-401, 2010
How does it work?
25. Strongest Correlate: Weight Gain (p=0.003)
“Explained” by three independent changes:
Adjust for:
Impact on
regression Residual p-value
Falling SI -40% <0.04
Falling Adiponectin -19% <0.02
Rising CRP -19% <0.02
All three -70% <0.29
Adipoiknes may directly influence the propensity for β-cells to fail.
Results
Predicting Changing β-cell Function
Xiang et al: Diabetes Care 33:396-401, 2010
27. Can we do anything to stop
progression to diabetes?
28. Feasibility and Mechanisms for Diabetes Prevention
Subjects: Prospective cohort of women Hispanic with prior
GDM (n=266)
Setting: GCRC-based study
Design: Interventional clinical trial with detailed physiological
measurements
Main Outcomes: diabetes rates and mechanisms for diabetes
prevention and beta cell preservation; pre-clinical
atherosclerosis
Investigators: Tom Buchanan, Anny Xiang, Ruth Peters, Stan
Azen, Howard Hodis, Wendy Mack
Support: Investigator-initiated pharmaceutical grants (Parke-
Davis and Takeda), NIH and ADA supplemental funding
Prevention Cohort
29. Preventing Type 2 Diabetes
Three Levels of Opportunity
Adipose
Tissue
Liver &
Muscle
Adipokines
Fatty Acids
Insulin Resistance
2
Insulin
Resistance
1
Obesity
Energy Balance
Negative Positive
Weight Loss
and Wasting
Fat
Accumulation
3
β-cell
Failure
Weak B-cells
Hyperglycemia
Robust B-cells
Hyperinsulinemia
TZDs
30. Overview of Design: Diabetes Prevention
TRoglitazone In Prevention Of Diabetes: TRIPOD Study
Buchanan et al: Diabetes 51:2796-2803, 2002
Hispanic women
with prior GDM
2000 - 20011995
Off
drug
Blinded Troglitazone
Blinded Placebo
TRIPOD Trial
Age: 34 yrs
BMI: 30 kg/m2
Fasting glucose: 98 mg/dl
2-hr Glucose: 154 mg/dl
HbA1C: 5.7%
OGTTs: Diabetes
IVGTTs:
Insulin Resistance
β-cell Function
32. On Trial
Off
Trial
Months after Randomization
FractionwithDiabetes
60%
40%
20%
0%
0 20 40 60
Placebo
12.1% per year
Troglitazone
5.4% per year
21% per yearn=40
3% per year
n=44
ivGTT
Masking?
TRIPOD Study: Post-Trial Washout
Buchanan et al: Diabetes 51:2796-2803, 2002
33. p=0.01 between groups
Baseline 8 Months Post-trial
Placebo (n=40)
0 2 4 6
MINMOD SI
AcuteInsulinResponse
(uU/mlxmin)
200
400
600
800
0
0 2 4 6
MINMOD SI
Troglitazone (n=44)
39% fall
Women without Diabetes during Trial
Stable
TRIPOD: Preservation of β-cell Function
Diabetes Prevention
Buchanan et al: Diabetes 51:2796-2803, 2002
34. Overview of Integrated Design
Off
drug
Open Label Pioglitazone
PIPOD Trial
2004
TRIPOD and PIPOD
2000 - 20011995
Off
drug
Blinded Troglitazone
Blinded Placebo
TRIPOD Trial
Diabetes
Diabetes
Open Label Troglitazone
Open Label Troglitazone OGTTs
IVGTTs
Xiang et al: Diabetes 55:517-522, 2006
35. 0 2 4 6 8
0
400
800
1200
1600
DispositionIndex
(SIxAIRg)
Years
Effect of Pioglitazone after Placebo
β-cell Function in TRIPOD+PIPOD
n=32
Pioglitazone
PIPOD
p=0.14
OffPlacebo
TRIPOD
p=0.003
Off
Xiang et al: Diabetes 55:517-522, 2006
36. 0 2 4 6 8
0
400
800
1200
1600
DispositionIndex
(SIxAIRg)
Years
Effect of Pioglitazone after Troglitazone
β-cell Function in TRIPOD+PIPOD
n=27
Troglitazone
TRIPOD
p=0.24
Off Pioglitazone
PIPOD
p=0.12
Off
Xiang et al: Diabetes 55:517-522, 2006
37. Type 2 Diabetes Prevention
Results of Recent Randomized Trials
Study Subjects Intervention Rel. Risk
*Similar β-cell protection with pioglitazone
Finnish DPS I.G.T. Lifestyle 58%
U.S. DPP I.G.T. Lifestyle 58%
XENDOS I.G.T. Orlistat 45%
Weight
Loss
Stop-NIDDM I.G.T. Acarbose 25%
Metformin 31%U.S. DPP I.G.T.Glucose
Absorption,
Production
TRIPOD Prior GDM Troglitazone
55%*
DREAM I.G.T. Rosiglitazone 62%
Fat-induced
Ins. Resist.
ACT NOW I.G.T. Pioglitazone 72%
38. Reducing body fat or mitigating its biological
consequences provides the best evidence for
disease modification (β-cell protection) in “pre-
diabetes”.
Diabetes Prevention Trials
Important Lesson
39. What is the mechanism for
diabetes prevention with TZDs?
40. Overview of Design: Diabetes Prevention
TRoglitazone In Prevention Of Diabetes: TRIPOD Study
Hispanic women
with prior GDM
2000 - 20011995
Off
drug
Blinded Troglitazone
Blinded Placebo
TRIPOD Trial
Age: 34 yrs
BMI: 30 kg/m2
Fasting glucose: 98 mg/dl
2-hr Glucose: 154 mg/dl
HbA1C: 5.7%
OGTTs: Diabetes
IVGTTs:
Insulin Resistance
β-cell Function
Buchanan et al: Diabetes 51:2796-2803, 2002
42. Resistant Sensitive
Insulin Sensitivity
InsulinOutput β-cell Protection in TRIPOD
Baseline
3 Months
Diabetes
6% / yr
Diabetes
0% / yr
(IVGTTInsulinArea)
(MINMOD SI)
Buchanan et al: Diabetes, 2002
Fasting Glucose
-7%
-3%
p=0.01
Fasting FFA
-2%
-8%
p=0.17
Triglycerides
-17%
-19%
p=0.60
Is it really “unloading” that counts?
43. -40 -20 0 20 40 60
0
5
10
15
Piogiltazone
Initial Reduction in Insulin Output*
(% of Basal)
DiabetesIncidence(%/yr)
TRIPOD and PIPOD
β-cell “Rest” and Diabetes Rates
Troglitazone
*change in IVGTT insulin area, in tertilesXiang et al: Diabetes 55:517-522, 2006
44. Some Attractive Mechanisms
“Toxic” Effects of β-cell Loading
Unfolded protein response
insulin
Amylin (IAPP)
Oxidative stress
Direct amylin toxicity
All may increase apoptosis and impair insulin secretion
46. Genetic Determinants of GDM and Type 2 Diabetes
Subjects: Mexican American GDM and control probands and
family members recruited from a variety of sources across Los
Angeles (n=~2200)
Setting: GCRC-base study
Design: Cross-sectional with longitudinal sub-study
Main Outcomes: associations between putative diabetes
genes and intermediate phenotypes for glucose regulation
Investigators: Tom Buchanan, Richard Watanabe, Anny
Xiang, Hooman Allayee, Jean Lawrence
Support: NIH R01s and ADA clinical research grants
Genetic Cohort
48. New Research Directions
Gestational Diabetes Study Group
Impact of GDM on obesity and glucose regulation in
offspring: Katie Page (CTSI K12 Scholar)
Enhancing compliance with follow-up for detection and
prevention of diabetes: Penina Segall-Gutierrez (CDC
research grant)
Bariatric surgery and β-cell preservation: Tom Buchanan,
Anny Xiang, Namir Katkhouda, Elizabeth Beale (planning)
Ethnic differences in glucose regulation after GDM: Anny
Xiang, Tom Buchanan (planning)
49. Prevention and Early Treatment of Type 2 Diabetes
One Clinical Strategy
Measure Glucose
Diabetes
Diet+Exercise
and Medication
Diabetic
Higher Risk
Diet and
Exercise
Impaired
Continue Diet
and Exercise
Stable
Glycemia
Medication
DevelopsD
iabetes
Lower Risk
Normal
Diet and
Exercise
Annual
Follow-up
Prevention
Treatment
High Risk Individual
50. β-cells fail when they are exposed to obesity and
insulin resistance.
“Overload” appears to be an important mechanism
contributing to β-cell failure.
Unloading can preserve β-cell function and prevent
diabetes. Weight loss and TZDs are the best
approaches we have right now – but not perfect.
We have to do something big about obesity if we
are to have any profound impact this problem.
Clinical and translational research is a great way to
do science with your friends and have a real impact
on human health.
Summary
This slide summarizes the study group, intervention and risk reduction for recent randomized trials of diabetes prevention.
Note that some of the differences in absolute risk reduction may be due to differences in length of treatment (longer = greater absolute risk reduction if diabetes rates are diverging in study groups)
Also not that the interventions have not been tested in combination in any study to date and that only metformin and lifestyle have been compared directly in the same study (the U.S. DPP)