EnVision CF Multicenter Study of Glucose Tolerance in Cystic Fibrosis

A. Hypotheses and Specific Aims:

Cystic Fibrosis Related Diabetes Mellitus (CFRD) has been identified by the CF community as
one of the top ten priorities for CF research1. In CF, clinical decline due to dysglycemia
begins early, prior to the diagnosis of CFRD2 and increases pulmonary decline and mortality3.
However, current definitions of dysglycemia were derived from non-CF populations, and
unfortunately, there is presently no way to determine who, of those with CF and dysglycemia,
will experience morbidity, mortality or greatest risk for CFRD. However, the CF Center in
Milan found that measurable age and sex dependent variables on oral glucose tolerance testing
(OGTT) strongly predict β-cell failure 4- the primary driver of clinical decline. We propose
a multi-center trial to determine age and sex dependent reference for fsOGTT-derived
estimates of β-cell function and insulin sensitivity in the US CF population, and to
investigate the relationship of these measures to CF-relevant clinical outcomes as well as
exploring potential mechanisms that may be driving beta cell functional decline. Furthermore,
we will collect these data in the new age of cystic fibrosis transmembrane conductance
regulator (CFTR) modulator therapies, to define a model that will be most relevant for our
current generation of CF patients.

Specific Aim 1: To determine age and sex dependent reference values from OGTT and
OGTT-derived estimates of β-cell function and insulin sensitivity in a diverse population of
475 non-diabetic children and adults with CF, at 4 centers with EnVision Emerging Leaders as
site-PIs.

Hypothesis: Much like standard growth charts with means and standard-deviation based
percentiles, fsOGTT values (glucose, insulin, c-peptide) and the resultant fsOGTT-derived
estimates of β-cell function (beta cell sensitivity to glucose, oral glucose insulin
sensitivity) will fall along an age and sex-dependent continuum which will be consistent
across CF patients at four different US centers.

Specific Aim 2: To determine the relationship between OGTT reference values and estimates of
β-cell function and insulin sensitivity to clinically relevant outcomes.

Hypothesis: The fsOGTT-derived values of insulin, glucose and C-peptide in combination with
estimates of β-cell function (beta cell sensitivity to glucose, OGIS) will correlate with
important markers of clinical status (FEV-1, BMI Z score, number of pulmonary exacerbations
over the past 12 months) in an age and sex dependent manner.

Specific Aim 3 (exploratory): To gather additional measures in a subset of this large cohort
to illuminate the relationship between normal ranges of beta cell function (as determined
above) to possible mechanisms driving abnormal beta cell function, including: abnormal lean
body mass, abnormalities of hormones that regulate beta cell function (specifically incretins
(glucagon-like peptide-1 (GLP-1), glucose-dependent insulinotropic polypeptide (GIP)), and
islet hormones (glucagon, pancreatic polypeptide (PP)), abnormalities of counter-regulation
to abnormal insulin secretion (catecholamine levels), and free-living glucose measures
collected by continuous glucose monitoring. Additionally, stored serum from 475 patients with
exceptionally well characterized metabolic profiles provides an unprecedented opportunity to
help inform future research questions.

Hypothesis 3a: Lower lean body mass as assessed by DXA is associated with beta cell glucose
sensitivity in an age and sex dependent manner.

Hypothesis 3b: Patterns of islet hormone and incretin secretion are age and sex dependent and
β cell glucose sensitivity will correlate with PP and GIP, but not with glucagon and GLP-1
Hypothesis 3c: Reduced β cell glucose sensitivity is associated with higher rates of reactive
hypoglycemia with inappropriate catecholamine and glucagon response during OGTT.

Hypothesis 3d: identifiable CGM measures (ex. Peak glucose, time spent >200 mg/dl, standard
deviation) will correlate with model derived outcomes specifically β cell glucose sensitivity
in an age and sex-dependent manner

B. Background and Significance:

Cystic Fibrosis Related Diabetes (CFRD) is an independent cause of increased rates of
premature death in people with cystic fibrosis (CF)3. Because of this, the CF community has
recognized CFRD as one of their top ten research priorities1. Importantly, abnormalities of
blood glucose metabolism (dysglycemia) are present in CF likely from birth5 and cause
increasingly rapid lung function decline and loss of healthy body mass years before the
diagnosis of CFRD is made2. However, it is unclear at what point individuals might benefit
from treatment of dysglycemia, as the treatment itself can be an additional burden.
Furthermore, with the increasing use of cystic fibrosis transmembrane conductance regulator
(CFTR) modulators, the natural history of CFRD development and progression is likely to
change. Thus, it is imperative that we better understand which CF patients with dysglycemia
are at the greatest risk of excess mortality due to dysglycemia and progression to CFRD. The
primary defect leading to dysglycemia in CF is inadequate insulin secretion secondary to
abnormal function of the beta cell6. Importantly, understanding beta cell dysfunction is
essential to addressing every form of diabetes mellitus, not just CFRD. However, the gold
standard tests for beta cell function (hyperglycemic clamp and tracer studies) are complex
and therefore they are not easily utilized in large-scale studies. Due to this, there has
been extensive effort on the part of many centers to develop methods to assay beta cell
function from more readily accessible data points and, ideally, clinically useful ones. Dr.
Andrea Mari, from the Institute of Neuroscience in Italy, has developed a model, which can
derive multiple indices of beta cell function from frequently sampled OGTT, with excellent
concordance to the gold standard hyperglycemic clamp. This is especially relevant in the CF
population where OGTT testing is the recommended modality for diabetes screening. Dr. Mari's
model has been well-validated in normal subjects7, subjects with type 1 and type 2 diabetes
mellitus and in subjects with pre-diabetes8 and is an accepted alternative for the assessment
of beta cell function in large studies where clamp protocols are difficult or impossible9.
Importantly, Dr. Alberto Battezzatti, a pediatric endocrinologist at the CF center in Milan,
Italy, with a long dedicated career interest in CFRD, has expanded Dr. Mari's modeling to the
CF population. Dr. Battezzati has shown that there are strong relationships between the
insulin, glucose and C-peptide levels measured during 3 hour OGTT testing and the age and sex
of CF subjects4. Our goal, in concert with Dr. Battezzati and colleagues is to develop normal
ranges for these measures or, in essence, a "growth chart" for the OGTT in CF patients that
can accurately predict beta cell dysfunction and eventual progression to CFRD. Next steps are
to simplify the model to so that 2-3 discrete samples will accurately predict progression to
CFRD and morbidity in CF subjects.

We propose a collaboration with Dr. Battezzati and the Italian CF centers. We will validate
the Italian model in a large and diverse US population, and correlate it to specific clinical
outcomes relevant to morbidity and mortality in CF, which will allow for development of
predictors in the model specific to US subjects and ensure clinical relevance to our
patients. Furthermore, we will collect these results and follow patient outcomes in this new
era of CFTR modulator therapy, developing a model that will be most relevant for our current
generation of patients with CF. We will establish age and sex-dependent "normal ranges" for
glucose, insulin and c-peptide values from OGTT as well as determinants of beta cell function
derived from OGTT. Establishment of normal ranges of these measures will allow the
determination of which patients with CF are the lost abnormal and eventually determine who,
within a given glucose tolerance category in CF will experience clinical compromise and
progress to CFRD. Our long-term goal will be to develop from these normal ranges, including
normal ranges for indices of beta cell function and insulin resistance, the most predictive
time points and assays to develop a simplified model that will optimally predict risk of
clinical decline in a CF patient of a given age and sex. The ideal model will consist of 2- 3
discrete time points to accurately predict clinical deterioration.

All four of the investigators on this grant have established or are developing a long term
career focus on Cystic Fibrosis Related Diabetes Mellitus and dysglycemia in Cystic Fibrosis.
This project represents a significant opportunity to characterize a large cohort of subjects
with well-defined dysglycemia and beta cell dysfunction in collaboration with an
equally-large international cohort with matching data sets. It represents an unprecedented
opportunity to fully explore the depth and breadth of dysglycemia in CF as well as to
investigate the impact dysglycemia has on the health of people with CF. The long term value
of this project is also enhanced by the establishment of a biorepository of stored samples
that will be obtained from this cohort. This will allow the ability in the future to query
these samples to further investigate the natural history and pathophysiology of CFRD, and
potentially biomarkers with in a large cohort with clearly and accurately defined beta cell
function. Our group has unique assets to complete this task as we represent four unique CF
centers; all with dedicated CF endocrinologists recognized as Emerging Leaders in CF
Endocrinology by the EnVision program, with coordinated mentorship though that program by the
leading expert in the field of CFRD, Dr. Antoinette Moran.

C. Preliminary Studies:

Currently the recommended screening test for CFRD is the oral glucose tolerance test
(OGTT)10. Although abnormalities on this test predict long term outcomes in CF, the
diagnostic endpoints we currently use are based on data from other adults with type 2
diabetes and are not specific to CF. There is emerging evidence that CF patients who do not
meet formal criteria for glucose abnormalities by OGTT are still experiencing clinical
decline11,12. Therefore, there remains a knowledge gap regarding how we can best predict
clinical decline. Given that the primary pathology in CF is insulin deficiency, it is
reasonable to extrapolate that measures of beta cell function would be the best indicators of
outcomes in CF patients. Typical simplified methods of assessing beta cell function (such as
HOMA-B) utilize primarily fasting measurements, which are inappropriate in CF where fasting
abnormalities develop after the onset of frank CFRD and long after clinical decline is
established.

Dr. Andrea Mari has extensive data showing the effectiveness of his model in evaluating beta
cell glucose sensitivity (the ability of the beta cell to respond to respond to changes in
glucose concentration)7,13,14. The OGTT itself has poor reproducibility, however beta cell
glucose sensitivity as measured from frequently sampled OGTT has significantly better
reproducibility, with a coefficient of variation around 15-20%7,15. This model of beta cell
glucose sensitivity is an extremely accurate parameter to assess glucose control and has been
found to be a significant predictor for the development of type 1 and type 2 diabetes in
non-diabetic and pre- diabetic populations. This model allows, as well, measurement of first
phase insulin secretion (termed "rate sensitivity" - which is directly comparable to the
acute insulin response obtained from the gold standard IVGTT)16. This is significant, as loss
of first phase insulin secretion is the initial abnormality in the progression to CFRD. This
model is highly reproducible, and has also been studied longitudinally15,17 (see figure 1).

Preliminary data -islet and incretin hormones:

Dr. Larson Ode has devoted her career to understanding the origins of dysglycemia in CF. Her
work has upended previous dogma, finding that up to 39% of CF children 3 months to 5 years of
age may have dysglycemia18. However, we do not yet know what underlying mechanisms drive
dysglycemia in CF. In our small dataset (n=25 CF, n=9 control), 9 of 25 CF subjects had
dysglycemia (all controls were normal). Stored blood was assayed for islet and incretin
hormones to interrogate the difference between children with and without dysglycemia. CF
children with abnormal glucose tolerance had higher GIP levels (p<0.03) and a trend toward
lower GLP-1 levels compared to normal CF subjects (p= 0.06). All CF subjects had abnormally
low PP levels (p<0.01). Glucagon levels were uninformative (unpublished data- see Figure 1).
This implies that incretin and islet hormones may play a role in dysglycemia in CF. However,
data from larger and older cohorts are needed to fully understand incretin and islet hormone
patterns in people with CF. This proposed project will establish a biobank which will allow
the study of age based progression in biomarkers of islet dysfunction throughout a diverse
cohort including a large age-range of subjects. The deep understanding of CF typical beta
cell function changes with age that will be established by this cohort will allow greatly
improved understanding of the progression of dysglycemia in CF.

Experience with CGM:

Dr. Chan currently has a CFFT Shwachman grant to investigate the role of CGM in CF. She also
has previously used CGM in other studies, enrolling over 100 obese youth to undergo OGTTs and
wear CGM in order to better understand simpler measures of prediabetes and type 2 diabetes
screening in the obese adolescent population19,20. Although her CGM in CF study is currently
active, recruitment will be ending soon and will not overlap with the timing in this proposed
study. The present proposal, with its target enrollment of 475 individuals with CF, offers an
unprecedented opportunity to further our understanding of CGM and its relationship with the
proposed model and its novel age- and sex-specific outcomes

Experience with hypoglycemia research:

Dr. Moheet has extensive experience in diabetes research with special expertise in the gold
standard tests for beta cell function (insulin clamp studies). His research utilizes methods
relevant to the study of in vivo metabolism in human subjects, particularly use of both
hyperglycemic and hypoglycemic insulin clamps methodologies. His research has examined the
underlying pathogenesis in development of defective counterregulatory response to
hypoglycemia in people with type 1 diabetes. Dr. Moheet's substantial background in
hypoglycemia research would be invaluable in the conduct of this study.

Fasting and reactive hypoglycemia after meals is common in CF patient without diabetes21.
Reactive hypoglycemia can be particular bothersome and difficult to manage in some patients
with CF. Hypoglycemia has also been frequently reported during OGTT in CF with reported rates
for 3 hour OGTT between 15-59%22,23. A recent systematic review noted that, however, that the
pathophysiology is poorly understood and highlighted the need for high quality studies to
examine the possible etiologies underlying reactive hypoglycemia in CF24. This proposed study
will address this knowledge gap by evaluating counterregulatory hormones including glucagon
and catecholamines during hypoglycemia and assessing symptomatic hypoglycemia via a validated
and structured questionnaire25. This proposed study will provide an unprecedented opportunity
to examine the possible mechanisms underlying reactive hypoglycemia in CF in a large cohort
of patients

Preliminary data-body composition:

Dr. Granados has previously investigated the association between body composition (% fat and
LBM by DXA) and measurements of insulin resistance (HOMA-IR), insulin sensitivity (Insulin
Sensitivity Index - Matsuda (ISIMatsuda)) and β-cell insulin secretion (Φtotal) in 15 CF
patients with pancreatic insufficiency between 12-24 years old (mean BMI 19.5 ± 2.87 kg/m2).
Her preliminary data showed that insulin resistance significantly correlated with weight
z-scores (HOMA-IR:r=0.48, p=0.003) and ISIMatsuda (r=-0.46, p=0.004). She also found a
positive correlation between Φtotal, (measurement of insulin secretion in response to
glucose) and fat % in the subjects with CF without diabetes (r=0.69, p=0.004). Moreover,
there was a trend in those with lower fat % and higher LBM to have better insulin sensitivity
(ISIMatsuda:r=-0.45, p=0.09).

D. Experimental Design and Methods:

Study Hypothesis Primary hypothesis: Much like standard growth charts with means and
standard-deviation based percentiles, fsOGTT values (glucose, insulin, c-peptide) and the
resultant fsOGTT-derived estimates of β-cell function (beta cell sensitivity to glucose, oral
glucose insulin sensitivity) will fall along an age and sex-dependent continuum which will be
consistent across CF patients at four different US centers.

Secondary hypothesis: The fsOGTT-derived values of insulin, glucose and C-peptide in
combination with estimates of β-cell function (beta cell sensitivity to glucose, OGIS) will
correlate with important markers of clinical status (FEV-1, BMI Z score, number of pulmonary
exacerbations over the past 12 months) in an age and sex dependent manner.

Exploratory hypotheses:

Hypothesis 3a: Lower lean body mass as assessed by DXA is associated with beta cell glucose
sensitivity in an age and sex dependent manner.

Hypothesis 3b: Patterns of islet hormone and incretin secretion are age and sex dependent and
β cell glucose sensitivity will correlate with PP and GIP, but not with glucagon and GLP-1
Hypothesis 3c: Reduced β cell glucose sensitivity is associated with higher rates of reactive
hypoglycemia with inappropriate catecholamine and glucagon response during OGTT.

Hypothesis 3d: Identifiable CGM measures (ex. Peak glucose, time spent >200 mg/dl, standard
deviation) will correlate with β cell glucose sensitivity in an age and sex dependent manner.

Screening and Enrollment The University of Minnesota CF center has 332 CF patients eligible
for this study. The University of Iowa has 144 eligible subjects. The University of Colorado
pediatric center has approximately 225 eligible subjects. Washington University in St. Louis
has 236 eligible subjects. Our experience from recruiting previous OGTT based studies in
these centers suggests 40-60% of CF subjects will agree to participate. We expect
approximately 160 subjects/year, collectively over all the sites, to reach at total of 475
subjects by the end of the 3 years of the study.

Experimental design and procedures:

Study Assessments:

Data to be collected at enrollment

1) Age, sex, date of birth, diagnosis of CF (genotype and sweat test results), any diagnosis
of diabetes mellitus, date of last hospitalization for pulmonary exacerbation and initiation
of any systemic steroid therapy. We will collect a list of the subject's current and active
medications from the parent (if the subject is a minor) or from the subject (if the subject
is an adult).

2) Medical Records: We will obtain consent to request the following information from clinic
and hospital medical records for participants, with HIPAA authorization:

a. medications, results of pulmonary function testing, specifically most recent FEV1%
predicted, FVC, CFTR genotype and seat testing results,. history of steroid use, number of
hospitalizations for pulmonary exacerbation in the last year (defined as admission for iv
antibiotics or steroid therapy), fecal elastase Data to be collected at Study visits

Frequently sampled 3-hour oral glucose tolerance test (OGTT):

1. Subjects are admitted to the clinical research unit

2. Fasting status is confirmed

3. An IV is placed for sequential laboratory draws. Topical lidocaine cream will be
provided for the IV start site if requested by the family.

4. The oral glucose solution (Glucola) at the dose of 1.75 g/kg is administered by mouth at
time zero.

5. Insulin, glucose, c-peptide, and additional serum to store will be obtained at 0, 15,
30, 60, 90, 120, 150 and 180 minutes. Additional blood will be stored for future
testing.

1. Each sample should require no more than 1-2 tbsp of blood. Sample amount will be
based on patient weight and will not exceed 3 ml/kg/day total blood withdrawn.

2. Samples will be processed, batched, and shipped to the University of Iowa for
analysis.

3. If the IV is lost at any point during the study, we will request permission from
the family to obtain samples by finger-stick or heel poke.

6. Anthropometrics at each visit.

a. Height on stadiometer for all subjects b. Weight on standing scale for all subjects.
c. These measurements to be used to calculate BMI and BMI z-score if applicable.

7. Additional data collection.

1. Medication list will be recorded and verified with the subject/parents at the start
of each visit.

2. From the medical record:

i. Last hospitalization, number of hospitalizations in the past year ii. Last systemic
steroid therapy (if ever) iii. Last Antidiabetic medication use (if ever) iv. Diagnosis
of diabetes- yes/no, date v. Fecal elastase values vi. CFTR mutation vii. Results of
sweat testing viii. Use of CFTR correctors/modulators (if any)- Starting date A subset
of the subjects will complete additional assessments evaluating the exploratory
endpoints. The goal is that 1/3 of the total subjects will participate in these
protocols.

DXA scan

1) A DXA scan will be obtained the date of the study visit or within a 2 week period from the
OGTT.

a. GE Healthcare (Madison, WI, USA) and Hologic Inc. (Bedford, MA, USA) are the 2 main
manufacturers. Although the DXA technology is similar for these two manufacturers, BMD and
body composition results differ because of proprietary calibration methods, algorithms to
body composition in every institution. For this multicenter clinical trial we plan to compare
the body composition results using generally accepted cross-calibration relationships
published for whole-body DXA30.

CGM

1) A blinded CGM Freestyle LibrePro (Abbott Diabetes Care Inc) will be placed on the back of
the upper arm. Participants will be asked to wear the CGMS for up to 10 days.

1. The CGMS does not require calibration, but participants will be asked not to wear the
monitor in water deeper than three feet, or to keep it submerged longer than 30 minutes.

2. Participants will also be instructed not to change any of their current dietary or
activity habits so that free-living glucose data are collected

3. They will also be asked to complete a simple log of their activity and dietary intake.

4. Subjects will be given the option of personally returning the CGM and log-sheet to the
CTRC or be provided a self-addressed, stamped envelope to return the device and
log-sheet at the completion of the recording period.

5. Using software that accompanies the CGM recorder, reports will be generated for each
subject providing a summary of the sensor glucose data.

6. These results will be converted into raw data in an excel spreadsheet and saved on a
password-protected, secure network server, via a local computer, in a locked office to
which only the investigators will have access.

i. These raw data include sensor glucose values obtained every 15 minutes during device wear.

g. We will review CGM tracings to analyze additional variables including: i. area under the
total CGM curve ii. number of excursions >140 mg/dl and >200 mg/dl iii. maximum sensor
glucose after first meal of the day, maximum sensor glucose, sensor glucose mean iv. standard
deviation and other measures of glycemic variability v. Measures of hypoglycemia specifically
% time spent <60 and <70mg/dl. The Freestyle Libre Pro system has been approved in Europe and
has been studied in children as young as 4 years of age for up to 14 days of sensor wear. It
does not display blood glucose information in real time, and participants will not be given
the capability to view the CGMS data during the study. This CGM does not require any
calibration or training and can easily be removed and returned Laboratory assays

1. Clinical research assays

a. Glucose, insulin, c-peptide- will be run on the via our protocol for research samples
clinical assays- utilize the Roche platform and are CLIA certified i. Total serum
insulin is measured via electrochemiluminescence immunoassay and utilizes the Roche
platform ii. Plasma glucose is measured via hexokinase iii. Plasma C-peptide is measured
via electrochemiluminescence immunoassay

2. Research only assays will be performed by Ms. Yi in the flow core at the University of
Iowa, all assays except glucose use the same platform, which allows multiple samples to
be run accurately on microliters of blood a. GLP-1 (active &total), GIP, PP, glucagon,
c-peptide, insulin i. HMHEMAG-34K | MILLIPLEX luminex xMAP Human Metabolic Hormone
Magnetic Bead Panel - Metabolism Multiplex Assay (Millipore Sigma, Burlington, MA, USA)
ii. No cross-reactivity is seen within the panel, the assays show high recovery
(97-105%), good precision (all assays have intra-assay coefficient of variation<10% and
inter-assay coefficient of variation <15%, except GLP-1 total which has an inter-assay
coefficient of variation of <20%) iii. Insulin an c-peptide are measured on this
platform as well to allow for quality control comparison with clinical samples

Sample size estimates/ Power calculation:

Primary outcomes:

Using 187 subjects from a single Italian center, the Italian group has previously defined the
75.0 quantile for various glucose tolerance and beta cell function parameters, including
beta-cell glucose sensitivity. As the goal of the present project is to develop the 5th to
95th quantiles, due to the clinical relevance of cut offs based on the 5th and 95th
percentile, sample size for the present project was calculated to estimate the 95.0 quantile
of a covariate-dependent quantile curve with a pre-specified precision. The primary outcome
variable of the quantile curve is beta-cell sensitivity to glucose and the predictor variable
is age. On the basis of the available data 4 we expect a linear relationship between
beta-cell glucose sensitivity and age with no heteroskedasticity. Therefore the minimum
sample size needed to estimate the 95.0 quantile of beta-cell glucose sensitivity -for-age
with a 90% probability for it to lie between the 90.5 and 97.4 quantiles (with respect to
symmetric tolerance evaluated on a logit scale) is 475 subjects31. For the other outcomes
(glucose, insulin, c-peptide) the minimum sample size is smaller than the 475 subjects needed
for primary endpoint of beta cell glucose sensitivity. The University of Minnesota CF center
has 332 CF patients eligible for this study. The University of Iowa has 144 eligible
subjects. The University of Colorado pediatric CF center has 225 eligible subjects.
Washington University in St. Louis has 236 eligible subjects. Historically at the University
of Minnesota and the University of Iowa, we have had excellent enrollment rates of CF
subjects in CFRD studies, so we expect about 40-60% of eligible subjects to enroll at some
point over the 3 year duration of the study.

Safety evaluation and quality control A formal DSMP will be in place for this project. Please
see attached DSMP. There will be no need for efficacy analysis as this is not an
interventional study.

Inclusion Criteria:

1. Age >/= 6 years

2. Diagnosis of cystic fibrosis

3. CF patients regularly attending the CF centers

4. Clinically stable in previous 3wks:

- absence of major clinical events including pulmonary exacerbations,

- no change in their habitual treatment regimen including introduction of
antibiotics or steroids in the past 3 weeks

Exclusion Criteria:

1. Diagnosis of type 1 diabetes, type 2 diabetes, or MODY

2. Organ transplantation

3. new diagnosis of CFRD in the past 6 months

4. antidiabetic treatment in past 6 mos (insulin or oral hypoglycemic agents)

-patients with previous CFRD diagnosis, but not currently taking
insulin/glucose-lowering medications for at least 6 months should be included

5. pulmonary exacerbation associated with systemic steroid requirement in the last 6
months

6. on CFTR corrector less than 6 months prior to enrollment