Progress Towards Closed-Loop Artificial Pancreas Systems in Diabetes

William V. Tamborlane, MD Professor of Pediatrics Yale University School of Medicine

“Bad Old Days” of Diabetes (Before 1980) • 1 or 2 insulin injections of NPH & regular insulin/day • Urine tests • Aggressive therapy unsafe and of unknown benefits • HbA1c 11-12% • Eye & kidney complications

Era of Intensive Treatment (1980’s)

• Blood Glucose Monitoring • HbA1c measurements • Insulin Pumps

1

DCCT Results Improved diabetic control with intensive vs conventional treatment decreased the risk of development and progression of early: – Retinopathy by 50-75% – Nephropathy by 35-55% – Neuropathy by 60% Three-fold increase in the risk of severe hypoglycemia DCCT Research Group, N Engl J Med 1993; 329:977-986

Technologic Advances Since the DCCT

• Insulin Analogs • Smart Insulin Pumps • Continuous Glucose Monitoring Systems

Why do we need an artificial pancreas? • Too many T1D patients fail to achieve target A1c goals • Severe hypoglycemia remains an ever present danger • Burden of care are extremely high and have increased with the translation of new diabetes technologies into clinical practice

2

Why not islet cell replacement therapies?

• Limited to small segments of population due to limitations in supply • Problems with rejection have not been overcome • They are not well suited for children with T1DM due to excessive morbidities related to immuno-suppression.

Essential elements of CL Systems Already Available

Control Algorithm

Continuous glucose sensor Insulin pump

All Three Elements Already Available Insulin Pumps Decades of experience with this technology • Now used by large numbers of youth with T1DM • Newer models better than ever •

Continuous Glucose Monitors •Commercially available for >10 years •Shown to be efficacious for open-loop therapy •Strong pipeline of improved devices

Controller Algorithms •PID •MPC •Fuzzy Logic

3

a)

GLUCOSE (mg/dl)

Medtronic ePID UCLA study 300

MEALS

SG

200 100

SUBJECT

0

b)

10

Suplemental Carbohydrate

5 1

10

100 delivery

concentration model fit

8

80 60

6 40

4 2

20

0

0

INSULIN (µ µU/ml)

c)

INSULIN (U/h)

12

d) Steil GM, et al. Diabetes. 2006;55:3344-3350.

Lessons Learned Exaggerated post-meal excursions and a tendency to late post-prandial hypoglycemia due to lags in: • Carbohydrate absorption • Increases in interstitial glucose concentrations • Insulin absorption from subcutaneous site

Excellent overnight control but lingering concerns re sensor accuracy

Possible Solutions Exaggerated post-meal excursions: • Hybrid, semi-automatic control with “priming” conventional pre-meal bolus to cover some of carbohydrate in meal Sensor error: • Set slightly higher than normal target glucose value (e.g. 120 rather than 90 mg/dL) to avoid nocturnal hypoglycemia

4

Reference Glucose Levels in CL vs Hybrid Control Glucose (mg/dl)

300 setpoint

Closed Loop (N=8)

meals

Hybrid CL (N=9)

200

100

0 6A

Noon

Full CL Hybrid

6P

MidN

6A

Noon

Mean

Daytime

Peak PP

147 ± 58 138 ± 49

154 ± 60 143 ± 50

219 ± 54 196 ± 52

6P

Weinzimer SA. Diabetes Care 2008; 31:934-939.

Percentage of glucose levels in or out of target range < 50 = 2 %

< 50 = 0 %

9

15

3

33 180

58

82

Closed-Loop

Home CSII Weinzimer SA. Diabetes Care 2008; 31:934-939.

P