Mixed Meal Test in Type 1 Diabetes on Insulin Pump Therapy: Optimization of Artificial Pancreas

BACKGROUND: Ideally, optimal treatment of T1DM should achieve near-normal glucose with no hypoglycemias and preserved quality of life. In the clinical arena, even the currently available most complex strategy, i.e. glucose sensor augmented insulin pump therapy, in which patients wear both a pump for subcutaneous insulin delivery (CSII) and a subcutaneous glucose sensor for continuous glucose monitoring (CGM), does not fully meet the patients' needs. Artificial pancreas (AP) or integrated closed loop control (CLC) algorithms -which take into account CGM readings and the effects of previous insulin infusions to continuously compute the amount of insulin dose to be administered- aim to minimize, in real time, glucose variability and prevent extreme glucose excursions. A number of research groups have been brilliantly working for years to develop a wearable AP to partially, or totally, free the patient (and/or the caregiver) from the burden of the open loop control. However, a perfect glucose control is still lacking, especially in post-meal glycemic excursions. Further insights might be gained by studies of the substrate/hormone responses to meals in T1DM patients on sensor augmented CSII, possibly compared with the gold standard of normal physiology, with special regard to the relative role played by a number of factors of the glucose-insulin pump system in controlling glucose levels.

Metabolic control analysis (MCA) is a theoretical framework which aims at quantitatively describing metabolic systems and at gauging distribution of control. MCA quantifies the role of each component of a system (e.g.: absorption rate of the fast insulin analogue, carbohydrate absorption through the gut, etc.) in controlling a variable of interest, most notably fluxes or concentrations, of the system itself, by computing the scaled coefficient of controls (CCs). CCs typically range from -1 to +1, and the greater their absolute value, the stronger is the control. Application of MCA to T1DM patients on CGM/CSII could quantify the individual role played by each components of the system, and may help in understanding and potentially overcome the hindrances to achieve stable and tight glucose control and to further develop CLC algorithms for optimal management of the disease.

AIM: To investigate characteristics and determinants of the glucose-insulin pump system in patients with T1DM with CGM with particular emphasis on the system level (i.e. glucose CCs). We will perform single mixed meal tolerance tests as well as 24-hour studies in T1DM patients and in healthy controls. We will apply mathematical modelling tools developed by us and will create virtual patients and virtual controls from this database to gain further insights into the alterations of the glucose-insulin pump system in patients.

SPECIFIC AIMS:

• AIM 1: To outline and compare the features of the glucose-insulin system in T1DM patients and in healthy controls during breakfast meals of different sizes
• AIM 2: To test the reproducibility of model assessed glucose-insulin pump system in T1DM patients
• AIM 3: To outline and compare the features of the glucose-insulin system in T1DM patients and in healthy controls during 24-hour periods

HYPOTHESIS: We hypothesize that in T1DM patients on CGM and CSII:

1. there are distortions in the distribution of plasma glucose control, as assessed by MCA, when compared to healthy individuals;
2. the distribution of control is different between plasma glucose and glucose derived from presently used CGM, which is the putative signal to be used in CLC algorithms of a wearable AP;
3. both the above mentioned putative alterations need be taken into account to develop effective and safe CLC algorithms.

METHODS: This research is articulated in three distinct protocols. All subjects will be on an isocaloric, maze and cane sugar free diet for one week before study. All T1DM patients will be on CGM/CSII and will be recruited from the patient population attending our Outpatient Insulin Pump Therapy Clinic. Each patient will continue to use his/her own pump device, insulin analogue and glucose sensor. All studies will start at 07:00 AM after an overnight fast (≥ 10 hours). All blood samples will be collected in refrigerated tubes and promptly spun at 1500 g; plasma will be separated and stored at -80° C. According to the planned use of plasma, tubes for blood samples will contain ethylenediaminetetraacetic acid (EDTA) for anticoagulation±antiglicolytic agent±serine protease and DPP4 inhibitors.

-PROTOCOL 1: Investigation of the glucose-insulin system during two mixed meal tests in T1DM patients and in healthy controls.

10 T1DM patients and 10 healthy age gender and BMI matched control subjects will be recruited. Patients will be on their usual insulin pump therapy, and the rate of the basal insulinization will be kept constant from 02:00 AM until the end of studies. All subjects will be studied on 3 separate occasions at 1-2 weeks intervals. In patients, study 1 will be a standard 120 min euglycemic insulin clamp. Timed blood samples will be collected to measure insulin levels at 10'-30' intervals. Plasma glucose will be measured at bedside by a YSI Glucose Analyzer; subcutaneous glucose will be monitored by CGM. In controls, study 1 will be a frequently sampled IVGTT of 240 min of duration. Timed blood samples (n=27) for glucose, C-peptide and insulin will be collected. Studies 2 and 3 will be identical in both patients and controls. They will be performed in randomized order and will last 360'. After a 60 min baseline period, at time 0' subjects will ingest either a 320 kcal (study 2) or a 640 kcal (study 3) meal. Meal composition (%Kcal: 53.3% carbohydrate, 28.2% lipid, 18.5% protein) will be Indian corn polenta plus parmesan cheese. Patients will program their pump to inject at time 0' a meal insulin bolus dose, calculated as usual. Blood samples (n=20) will be collected at timed intervals to measure plasma glucose, 13C/12C-glucose ratio, insulin, glucagon and C-peptide (the latter one only in control subjects). At a lower time frequency, triglycerides, non-esterified fatty acids (NEFA), active glucagon-like peptide-1 (GLP1) and total gastric inhibitory peptide (GIP) will also be measured. Subcutaneous glucose will be monitored by CGM only in patients. Data derived from IVGTT in controls will take the role played by the insulin clamp data in T1DM patients. Past experiments and also unpublished observations in our laboratory have shown that similar values of insulin sensitivity are obtained with IVGTT and insulin clamps in healthy people. The other difference is that, instead of the insulin pump, a beta cell secretion minimal model will be used, as previously described. The availability of 12C/13C-glucose ratio determinations will enrich further this picture, allowing to compute the contribution of meal carbohydrate to total plasma glucose and, by difference, the contribution of endogenous (primarily liver) production to total glucose. This data enrichment will result into further model development at step 3 (in both controls and patients) and will allow to split total insulin action into its two components, the one acting on endogenous glucose production and the one acting on glucose utilization. The former will be relevant to interpret the glucagon and, possibly, also the active GLP1 data.

With this strategy, it is possible to compute all relevant CCs of G(t), for both plasma and CGM values. Phenotypic results are layered in three levels of ascending complexity:

1. Model-free measurements: meal substrate and hormone levels, with particular regard to glucagon and active GLP1 for their action on liver glucose production and to total GIP, as a regulator of fat cell metabolism;
2. Model-derived estimates of single traits of the glucose/insulin system: they include, but are not limited to, glucose-sensitivity (SG), insulin sensitivity (SI), carbohydrate transit time to reach systemic circulation, mean transit time of insulin to reach systemic circulation from the subcutaneous depot (only patients);
3. Model-derived assessment of system properties of the glucose-insulin system (controls) and of the glucose-insulin pump system (T1DM) patients: they include all the glucose (also CGM glucose in patients) coefficients of control.

PROTOCOL 2: Reproducibility of the assessment of the glucose-insulin system during a single mixed meal test in T1DM patients.

10 T1DM patients on insulin pump therapy and subcutaneous continuous glucose monitoring (CGM) will be recruited. The study protocol is identical to protocol 1 with the following exceptions: 1. in 5 T1DM patients study 2 and study 3 will be two repeated 320 kcal meals; 2. in 5 T1DM patients study 2 and study 3 will be two repeated 640 kcal meals. The phenotypic results will be layered as in protocol 1 for reproducibility analysis.

-PROTOCOL 3: The glucose-insulin system during a 24-hour period in T1DM patients and in controls.

10 T1DM patients and 10 healthy age gender and BMI matched control subjects will be recruited. All subjects will be studied on 2 separate occasions at 1-2 weeks intervals. Both in patients and in controls, study 1 will be identical to protocol 1 (i.e. an insulin clamp and an IVGTT, respectively). In study 2 all subjects will be admitted to the Clinical Research Center (CRC) on the morning of the study (07:00 AM) and will stay there for its entire duration (24 hours). Subjects will be allowed to move freely within the CRC, but they will not be engaged in any physical exercise. Patients will continue their usual insulin therapy, including meal insulin dose calculation and injection, throughout the study. At 08:00 AM, 13:00 and 19:00 subjects will ingest three mixed meals (200 kcal/m-2 of Body Surface Area (BSA), 400 kcal/m-2 BSA and 400 kcal/m-2 BSA, respectively; diet composition: 54% carbohydrate, 28% lipid, 18% protein). In 6 patients and in 6 controls, the 13:00 meal will be Indian corn polenta plus seasoned Italian DOP parmesan cheese. In the other 6 patients and 6 controls, the 19:00 meal will be Indian corn polenta plus seasoned Italian DOP parmesan cheese. Timed blood samples (n=62) will be collected. Plasma glucose, insulin, glucagon and C-peptide (the latter one only in control subjects) will be measured at all time points. At a lower time frequency, triglycerides, non-esterified fatty acids (NEFA), amino acids, active GLP1 and total GIP will also be measured. In those subjects ingesting the polenta meal at 13:00, the 13C/12C-glucose ratio will be measured from 12:00 until 24:00. In those subjects ingesting the polenta meal at 19:00, the 13C/12C-glucose ratio will be measured from 18:00 until 07:00 the morning after. Subcutaneous glucose will be monitored by CGM only in patients. The polenta meal will be used only in one meal per day to avoid loss of interpretability of 13C/12C-glucose ratio data, owing to ongoing glucose carbon recirculation/recycling through glycogen/gluconeogenesis. The behaviour of the glucose/insulin system at breakfast time is already explored in Protocol 1. Results will be layered and analyzed as described in Protocol 1.

EXPECTATIONS: The combined analysis of insulin clamp and mixed meal test (MMT) data will allow to build a comprehensive model of the glucose insulin system during a MMT in each subject, which will thereby work as an in silico virtual patient, according to a well established methodology developed in our laboratory. The collection of virtual patients with type 1 diabetes undergoing a MMT will form the in silico biobank derived from this study. Virtual patients will be used to carry out MCA to compute CCs. Furthermore, the incretin hormone, glucagon and substrate response to a mixed meal will be quantified, allowing the putative identification of further modifiers of the glucose-insulin system. This database will be instrumental in devising a control algorithm able to guarantee a normal glucose regulation during a mixed meal in patients with type 1 diabetes with CGM on insulin pump therapy.