Comparison of Metabolic Responses To The Mixed Meal Tolerance Test Versus The Oral Glucose Tolerance Test After Successful Clinical Islet Transplantation

minute glucose concentration following a 75g-OGTT (OGTT 120 ) ≥11.1 mmol/L and their diagnostic accuracy. Studies with OGTT 120 ≥11.1 mmol/L (n=5) had diminished C-peptide: glucose, greater integrated glucose and diminished insulin: glucose area under the curve (AUC) ratios (0-120 minutes) and disposition indices; all p<0.05, contrasting with MMTTs where no difference in the 90-minute glucose concentrations, C-peptide:glucose, integrated glucose, C-peptide and C-peptide: glucose AUCs (0-90 mins) was seen; all p>0.05. A 90-minute MMTT glucose concentration ≥ 8.0 mmol/L demonstrated a sensitivity and specificity of ≥ 80% for the diagnosis of OGTT 120 ≥11.1 mmol/L ; area under ROC curve (mean±SEM) 73±13%. A 90-minu te MMTT glucose ≥ 8.0 mmol/L, identifies islet transplant recipients who may require closer monitoring for graft dysfunction.


Introduction
Islet transplantation (ITx) is an effective treatment for patients with Type 1 diabetes with frequent, severe hypoglycaemia associated with impaired awareness of hypoglycaemia even in the absence of completely normal glucose tolerance or

Accepted Article
This article is protected by copyright. All rights reserved.
insulin independence (1)(2)(3)(4)(5). Intrahepatic ITx can achieve short-term insulin independence in almost all cases. Factors promoting ITx success include islet number (1,6) and purity, (7) but long term insulin independence is difficult to maintain (2,8,9). Attrition in graft function is well recognised although poorly understood. Mechanisms for the diminished graft function observed with time may include autoimmune and alloimmune processes and an intrahepatic environment that is toxic as a consequence of exposure of islets to high concentrations of immunosuppressive drugs; the relatively low oxygen concentration in the liver may also play a role (10). Monitoring of graft function (11) is therefore of importance in the follow up care of patients to identify individuals who may require metabolic or immunologic support to prevent further graft loss.
Measures of beta cell function may be made directly from circulating C-peptide concentrations (12) after stimulation with arginine (13) and glucagon tests (14) or indirectly using surrogate measures, for example using continuous glucose monitoring systems (15,16). Recently it has been demonstrated that a measure of ITx engraftment may be derived from a fasting C-peptide measurement (11) but islet transplantation programmes across the world have long recognised stimulated Cpeptide measurements as an appropriate primary outcome measure (2,4,17,18).
The most used method for stimulating C-peptide response in islet transplant recipients in clinical settings is the mixed meal tolerance test (MMTT) as it is highly reproducible and represents a robust but physiological stimulus for C-peptide secretion with a lower risk for hyperglycaemia, because a smaller load of glucose is used, as compared to an OGTT (19). In the MMTT, a liquid meal is ingested in the fasting state with timed measurements of C-peptide and other metabolites postprandially. In islet transplantation programmes, the post-prandial C-peptide

Accepted Article
This article is protected by copyright. All rights reserved.
concentration at 90-minutes is taken to reflect the peak stimulated circulating Cpeptide concentration and is interpreted in the context of the glucose concentration at this time point (20).
The purpose of metabolic testing with MMTT after islet transplantation is to assess graft function rather than to define glucose intolerance or recurrence of diabetes. This is in contrast to the use of OGTT to diagnose degrees of glucose intolerance including impaired fasting glucose, impaired glucose tolerance and post-transplant diabetes in other transplant settings (21), using the same thresholds defined by the WHO (plasma glucose ≥11.1 mmol/L (200mg/dL) 2hours after a 75g oral glucose load) to diagnose diabetes in the general population (22)

Accepted Article
This article is protected by copyright. All rights reserved.

Research Design and Methods
In the original Edmonton Protocol, beginning in 1999, MMTTs were performed routinely before and at 3, 6, 12, and 18 months, and OGTTs performed at 6 and 12 months, post-transplant (18). Subsequently, OGTTs were performed infrequently since patients disliked exposure to large glucose loads and generally only in insulin independent subjects. We analysed data from metabolic studies performed between 1999 and 2003 in subjects who had received their first islet infusion before 2002. The induction and maintenance immunosuppression received by subjects reflects the original Edmonton Protocol (daclizumab at induction and maintenance with sirolimus (8-10 ng/ml) and tacrolimus (4-6 ng/ml)).

Participants
Data from thirteen insulin-independent islet transplant recipients from Edmonton, with stable graft function, defined as capillary blood glucose readings <10mmol/L (tested 4 times per day over the previous 1 month), without exogenous insulin therapy or oral hypoglycemic agents with HbA1c <7% (53 mmol/mol) and normal renal function (eGFR>60ml/min) and who had undergone paired MMTTs and OGTTs are presented.

Metabolic studies
Three days before all studies, participants were asked to consume 250 g carbohydrates per day and abstain from alcohol or strenuous exercise, after which participants were studied in the Clinical Research Facility after an overnight fast of 8-

Accepted Article
This article is protected by copyright. All rights reserved.
10 hours. The MMTT and OGTT tests took place on separate study days. The order of the paired tests were randomised.
Height was recorded to the nearest 0.5cm, weight to the nearest 0.1 kg (SECA 761 scales) and information regarding medications over the previous one week recorded.
All participants had a 44-mm, 20-gauge cannula inserted in the left forearm for venous blood sampling. Participants acclimatised for 30 minutes prior to the ingestion of glucose or the mixed meal as previously described (27). The participant remained seated for the duration of the test.
75g OGTT: Sampling for glucose, C-peptide and insulin were done at baseline and then following ingestion of 75g glucose at time 30, 60, 90 and 120 minutes.
All samples were centrifuged at 3000 rpm for 15 minutes at 4°C, separated and the plasma frozen at -70°C until analysis.

Accepted Article
This article is protected by copyright. All rights reserved.
Biochemical assays were measured in duplicate and concentrations determined by the glucose oxidation method. C-peptide and insulin concentrations were measured using commercial kits (Roche Elecsys, Roche Diagnostics, Indianapolis, IN; and Pharmacia, Uppsala Sweden respectively) with lower limit of assay sensitivities of 0.02 nmol/L and 0.35 nmol/L respectively (27).

Ethical approval
Participants provided written informed consent and the study was approved by University of Alberta Health Research Ethics Board, and conducted in accordance with the principles endorsed by the Declaration of Helsinki and the Declaration of Istanbul.

Data Analysis
Measures of insulin resistance were derived from the HOMA-IR (28) and the Matsuda index (29) as measures of hepatic insulin resistance and whole body insulin sensitivity respectively (30). The insulinogenic index was derived as a surrogate measure of insulin secretion (31,32)

Accepted Article
This article is protected by copyright. All rights reserved.

Statistical analyses
Statistical analyses were performed using STATA version 15 (Stata Corporation, TX, USA). Descriptive statistics are expressed as mean±SEM and median(IQR) as appropriate. Data sets were compared using unpaired t tests, Mann Whitney U tests and ANOVA analyses with post-hoc testing. Comparisons between intra-individual OGTTs and MMTTs was compared by paired t-tests or Wilcoxon signed rank tests. an OGTT 120 ≥11.1 mmol/L with maximum sensitivity and specificity was also derived from the ROC curve analyses. In post-hoc analyses, ROC curves of the MMTT data were constructed using a DI <1 and the specificity and sensitivity of the 90-minute MMTT-glycaemic thresholds generated in the primary analysis were derived.
Statistical significance was set at 5%.

Results
Seventeen paired OGTTs and MMTTs were performed in 13 participants; 9 participants received one OGTT and one MMTT; four participants received two OGTTs and MMTTs ((mean±SEM) 6±1 months between each paired study).

Accepted Article
This article is protected by copyright. All rights reserved.
The time interval between first transplant and the metabolic test was (mean±SEM) 15±2 months. The paired MMTTs and OGTTs were performed a median of 2(1-2) days apart.
The personal data of the subjects is shown (Table 1) Table 2C). However, the change in C-peptide in relation to the glucose concentrations, did not differ as evidenced by the AUC C-peptide in relation to the glucose concentrations (Table 2C).

Accepted Article
This article is protected by copyright. All rights reserved.
When comparing studies with and without OGTT 120 ≥11.1 mmol/L there was no difference in HbA1c, beta or BETA-2 scores between the subgroups (Table 2A) (Table 2A). There was no significant difference in insulin sensitivity (HOMA-IR and Matsuda index). However insulin secretion (insulinogenic indices) and DIs were lower in the studies with OGTT 120 ≥11.1 mmol/L (Table 2A).
In contrast to the OGTT, there were no significant differences in the fasting and 90minute glucose and C-peptide concentrations or in the 90-minute C-peptide: glucose concentrations; AUCs for glucose and C-peptide between 0-90-minutes; and the Cpeptide: glucose ratio from 0-90-minutes during the MMTT between studies in those with or without OGTT 120 ≥11.1 mmol/L (Table 2B).
Individual glucose, C-peptide and insulin concentrations during the OGTT for each study are shown ( Figure 1C-E). The 120-minute glucose and C-peptide concentrations following the 75g-OGTT were significantly greater than the 90-minute MMTT glucose and C-peptide concentrations (Figure 2A and 2B) although there was no difference in the stimulated C-peptide: glucose ratios ( Figure 2C). Integrated

Accepted Article
This article is protected by copyright. All rights reserved.
When the analyses were repeated without the second observations in the 4 subjects, all results remained statistically significant.

Discussion
This study in insulin-independent islet transplant recipients with stable graft function and normal renal function, examined metabolic responses to a standardised oral glucose challenge and a mixed meal challenge. Not surprisingly there was a much greater increment in glucose after the larger glucose load of the OGTT, but both tests provided an equivalent estimate of graft function as judged by the integrated Cpeptide in relation to the circulating glucose concentrations. Furthermore, we have shown that a 90-minute MMTT glucose ≥ 8.0 mmol/L is equivalent to a 2 hour glucose ≥11.1 mmol/L after a standard 75g-OGTT. Of note no participants were on exogenous insulin or oral hypoglycaemics and therefore there was no confounding of

Accepted Article
This article is protected by copyright. All rights reserved. metabolic results. Even in this highly selected cohort of narrow segment of subjects with insulin independent islet transplant recipients with very good graft function (shown by beta and BETA-2 scores), stimulation tests can identify individuals with sub-optimal graft function whose grafts may benefit from some metabolic support.
The MMTT is a valuable tool used to quantify graft function, rather than to identify glucose intolerance. Nevertheless, post challenge glucose concentrations were significantly higher after OGTT than post MMTT which is consistent with the MMTT being a more physiological and less potent stimulus of insulin secretion and therefore a less stringent challenge to graft function. While the post MMTT glucose levels were not able to reliably discriminate between those with and without OGTT 120 ≥11.1 mmol/L, both OGTT and MMTT were useful stimuli for insulin secretion measured by AUC for insulin and C-peptide. Such physiological testing with a standardised MMTT which contains less than half the amount of carbohydrate compared to a 75g-OGTT, avoids unnecessary hyperglycaemia therefore exposing the islet graft to less metabolic stress, is the preferred option to assess graft function following islet transplantation and is widely adopted by islet transplantation programmes world-wide permitting comparisons between subjects or within subjects over time (4,27). Furthermore, such physiological testing has been adopted in new onset diabetes trials including TrialNet (19). A liquid meal is not truly physiologic however, and is associated with more rapid delivery of nutrients to the duodenum than after a solid meal (33). It does however avoid the confounding of delayed gastric emptying which might be anticipated in a cohort with long diabetes duration and high prevalence of diabetic neuropathy.

Accepted Article
This article is protected by copyright. All rights reserved. prevalence of impaired fasting glucose was seen. The mechanism is not known but diminished pulsatility of insulin secretion may play a role (37)(38)(39)(40)(41)(42). Such studies underline the importance of studying islet transplant participants distinct from other groups with diabetes and extrapolations of data from other subjects with diabetes including those who are C-peptide positive, may be inappropriate (43,44).
Of note the subjects selected for islet transplantation were insulin sensitive with normal body mass index (BMI) (6). Since insulin sensitivity was not measured by gold standard hyperinusulinaemic euglycaemic clamp studies, subtle defects in insulin sensitivity may have been missed. Certainly, immunosuppression with tacrolimus is recognised to induce insulin resistance and may contribute to insulin secretory deficits in some subjects; how this relates to the dose of immunosuppression as well as to their concentration is incompletely understood (45)(46)(47). In this study, subjects were receiving sirolimus and lower doses of tacrolimus (trough target 4-6 ng/ml) than is commonly employed when tacrolimus is combined with mycophenolate mofetil.

Accepted Article
This article is protected by copyright. All rights reserved.

Studies that have examined subjects with Type 1 diabetes with paired OGTT and
MMTTs have also demonstrated a 30% lower glucose concentrations with the latter test (48). Other studies comparing the standardised MMTT to a 75g-OGTT in subjects with a range of glucose tolerance who had repeated tests, demonstrated a linear correlation between the two tests at 120 minutes (49). This result concords with our observation of a positive association between the stimulated glucose values following the MMTT and the 75g OGTT, although in our study this just failed to meet statistical significance. The MMTT has been shown to be associated with lower glucose variability, fewer adverse symptoms and greater palatability versus the OGTT (49).
The primary analyses revealed a close relationship between the MMTT 90-minute glucose ≥8.0 mmol/L and glucose intolerance indicated by OGTT 120 ≥11.1 mmol/L.
Since the objective of metabolic testing is to assess graft function post-hoc analyses were also performed to explore insulin secretion using ROC curves constructed using the DI<1, a composite score reflecting defective insulin secretion in relation to the insulin sensitivity (30,50). Such analyses confirmed the close association of the 90-minute MMTT glucose with this further measure of graft dysfunction. Of note the HbA1c did not differ between islet transplant recipients with OGTT 120 ≥11.1 mmol/L versus those <11.1 mmol/L and therefore HbA1c was not examined further in secondary analyses. It is possible that recent onset of dysglycaemia could explain the similar HbA1c levels in subjects with and without diabetes, but more likely reflects the relatively low sensitivity of HbA1c for diagnosis of diabetes.

Accepted Article
This article is protected by copyright. All rights reserved.
Of note variability in fasting plasma glucose on the days of the tests was observed.
This did not reach statistical significance but intra-individual glycaemic variability of approximately 15% is well recognised even under standardised conditions as reported here (51).
Since a diagnosis of diabetes reflects inadequate insulin secretion to maintain euglycaemia these studies are important as they help identify a threshold where islet graft function may be sub-optimal. Although the MMTT does not clearly differentiate those with and without OGTT 120 ≥11.1 mmol/L, it does identify a corresponding threshold (90-min glucose ≥ 8mmol/L) associated with reduced DI. This threshold may identify subjects whose graft function should be monitored more closely; prompt efforts to minimise metabolic demands (perhaps through changes to nutrition and physical activity), or suggest consideration of therapies to support the graft or lower blood glucose levels (e.g. supplementary exogenous insulin, or other antihyperglycaemic agents). Delaying the institution of anti-hyperglycaemic therapies until overt hyperglycaemia, late in the spectrum of declining beta-cell function, will have limited ability to alter the natural history of beta-cell failure.

Accepted Article
This article is protected by copyright. All rights reserved.

Table 1. Personal Data of Subjects
Personal data in 13 islet transplant recipients undergoing n=17 paired OGTTs and MMTTs in study. Data shown is mean±SEM or median(IQR). p -statistical testing between islet transplant recipients with OGTT 120 <11.1 mmol/L versus OGTT 120 ≥11.1 mmol/L. *IEQ represents total number of islets received following transplants.**Neuropathy assessed clinically (autonomic and peripheral).

Accepted Article
This article is protected by copyright. All rights reserved. p -statistical testing between OGTT 120 <11.1 mmol/L versus OGTT 120 ≥11.1 mmol/L.