ArticlePDF Available

Glycemic Variability Assessment in Newly Treated Exocrine Pancreatic Insufficiency With Type 1 Diabetes

Authors:

Abstract and Figures

BACKGROUND: Thirty-nine percent of people with type 1 diabetes may have lowered pancreatic elastase levels, correlated with exocrine pancreatic insufficiency (EPI or PEI). EPI is treated with oral supplementation of pancreatic enzymes. Little is known about the glycemic impact of pancreatic enzyme replacement therapy (PERT) in people with diabetes. This article demonstrates a method of assessing glycemic variability (GV), glycemic outcomes, and other changes in an individual with type 1 diabetes using open-source automated insulin delivery (AID). METHOD: Macronutrient, PERT intake, and EPI-related symptoms were self-tracked; diabetes data were collected automatically via an open-source AID system. Diabetes data were uploaded via Nightscout to Open Humans and downloaded for analysis alongside self-tracked data (food, PERT). Glycemic outcomes, macronutrients, PERT dosing, and a variety of GV metrics following meals were evaluated for one month before and one month after PERT commencement. Breakfast was assessed independently across both time periods. RESULTS: In an n = 1 individual using an open-source AID, time in range was already above goal and improved further after PERT commencement. Glucose rate of change and excursions >180 mg/dL were reduced; mean high blood glucose index was reduced overall and more so specifically at breakfast following PERT commencement. CONCLUSIONS: GV can aid in assessing response to new-onset medications, as was demonstrated in this article for n = 1 individual with type 1 diabetes (using an open-source AID) after commencing PERT for newly identified EPI. GV may be useful for evaluating the efficacy of new-onset medications for people with insulin-requiring diabetes.
Content may be subject to copyright.
https://doi.org/10.1177/19322968221108414
Journal of Diabetes Science and Technology
1 –8
© 2022 Diabetes Technology Society
Article reuse guidelines:
sagepub.com/journals-permissions
DOI: 10.1177/19322968221108414
journals.sagepub.com/home/dst
Original Article
Introduction
Estimates indicate that around 39%1 of people living with
type 1 diabetes have lowered pancreatic elastase levels,
which can be indicative of exocrine pancreatic insufficiency
(EPI or PEI). EPI results in poor digestion of food which can
lead to malnourishment and can increase morbidity and mor-
tality due to osteopenia and osteoporosis.2 Treating EPI
involves oral supplementation of necessary pancreatic
enzymes alongside nutrient ingestion, known as pancreatic
enzyme replacement therapy (PERT).3 However, despite
PERT studies specific to the diabetes population, little is
known specifically about the glycemic impact of improved
digestion with the addition of PERT in people with diabetes
(PwD) who also have EPI, and PwD are typically cautioned
to watch for hypoglycemia or hyperglycemia.4 With the
advent of new diabetes technologies including continuous
glucose monitors (CGM), connected insulin pumps and
insulin pens, and automated insulin delivery (AID) systems,
it is now possible to answer questions that have previously
been left as gaps in the medical research.5 PERT, in particu-
lar, may be well suited to study with an assessment of changes
in post-meal glycemic variability (GV) in people with insu-
lin-requiring diabetes. Alongside analyzing diabetes data and
meal intake records from an individual with an open-source
AID for a period of time both before and after using PERT,
this article proposes a method for assessing GV and other
1108414DSTXXX10.1177/19322968221108414Journal of Diabetes Science and TechnologyLewis and Shahid
research-article2022
1#OpenAPS, Seattle, WA, USA
2CeADAR, Ireland’s Centre for Applied AI, University College Dublin,
Dublin, Ireland
Corresponding Author:
Dana M. Lewis, BA, #OpenAPS, Seattle, WA 98101, USA.
Email: Dana@OpenAPS.org
Glycemic Variability Assessment in Newly
Treated Exocrine Pancreatic Insufficiency
With Type 1 Diabetes
Dana M. Lewis, BA1 and Arsalan Shahid, PhD, MBA2
Abstract
Background: Thirty-nine percent of people with type 1 diabetes may have lowered pancreatic elastase levels, correlated
with exocrine pancreatic insufficiency (EPI or PEI). EPI is treated with oral supplementation of pancreatic enzymes. Little is
known about the glycemic impact of pancreatic enzyme replacement therapy (PERT) in people with diabetes. This article
demonstrates a method of assessing glycemic variability (GV), glycemic outcomes, and other changes in an individual with
type 1 diabetes using open-source automated insulin delivery (AID).
Method: Macronutrient, PERT intake, and EPI-related symptoms were self-tracked; diabetes data were collected automatically
via an open-source AID system. Diabetes data were uploaded via Nightscout to Open Humans and downloaded for analysis
alongside self-tracked data (food, PERT). Glycemic outcomes, macronutrients, PERT dosing, and a variety of GV metrics
following meals were evaluated for one month before and one month after PERT commencement. Breakfast was assessed
independently across both time periods.
Results: In an n = 1 individual using an open-source AID, time in range was already above goal and improved further after
PERT commencement. Glucose rate of change and excursions >180 mg/dL were reduced; mean high blood glucose index
was reduced overall and more so specifically at breakfast following PERT commencement.
Conclusions: GV can aid in assessing response to new-onset medications, as was demonstrated in this article for n = 1
individual with type 1 diabetes (using an open-source AID) after commencing PERT for newly identified EPI. GV may be useful
for evaluating the efficacy of new-onset medications for people with insulin-requiring diabetes.
Keywords
diabetes, enzyme supplementation, EPI, exocrine pancreatic insufficiency, glycemic variability, GV, pancreatic enzyme
replacement therapy, PERT
2 Journal of Diabetes Science and Technology 00(0)
related changes following the onset of new medication such
as PERT.
Methods
Data Collection
Data regarding meals from the post-PERT timeframe were
manually tracked by the individual in a spreadsheet, record-
ing PERT intake and quantity, partial macronutrient (fat and
protein) content, timing of the meal, and EPI-related symp-
toms as an outcome of the meal. Each meal or snack during
this time period thus included the date, time, macronutrient
estimate, enzyme supplementation type and quantity, the
timing of symptoms, and any other factors perceived to have
influenced the meal results.
In addition, as an open-source AID6 user, all diabetes data
from CGM, pump, and AID including blood glucose values,
insulin dosing data, manual carbohydrate entries, and algo-
rithmic decision making, were automatically uploaded to
Nightscout, an open-source remote monitoring platform.7
The same type of diabetes data were automatically recorded
for both pre- and post-PERT time periods.
Data Retrieval and Cleaning
The post-PERT meal-related data were provided as a single
.csv file for analysis. (No specific meal tracking data were
collected during the pre-PERT timeframe, other than carbo-
hydrate entries). To retrieve the diabetes-related data for
analysis, the Nightscout Data Transfer tool8 was used to pull
data from the selected time frames (pre- and post-PERT) and
store the data in Open Humans.9 The data were downloaded
and converted from JSON to CSV using the Unzip-Zip-
CSVify-OpenHumans-data.sh script.10 The resulting CSV
files (entries, containing BG data and timestamps; profile,
including profile data; treatments; and devicestatus, includ-
ing algorithmic decisions and insulin dosing data) were then
cleaned for analysis. In the process of cleaning the CGM
data, the BG data for the individual was first compiled into
one file (which was previously split during CSV conver-
sion), timestamps were cleaned and consistency was ensured
between the file types, and error codes were removed.
Data Analysis
Data were assessed separately for the pre- and post-PERT data.
The post-PERT data were collected during a four-week period
starting with the commencement of PERT in early January
2022, containing 13 975 glucose entries. Data were collected
by an adult (early 30s) female with newly diagnosed EPI hav-
ing an elastase level of 200 µg/g, discovered after more than 19
years of living with Type 1 diabetes. A similar, four-week time-
frame (also containing 13 975 glucose entries) was selected in
November 2021 for pre-PERT comparison purposes.
Since the data are a time series, we analyzed and plotted
autocorrelation (ACF) and partial autocorrelations (PACF)
of the sensor glucose values for 70 data points.
We then performed the following statistical and GV
analyses:
Visual analysis of glucose profiles and calculation of
glycemic outcomes including time in range (TIR,
70-180 mg/dL), above range (>180 mg/dL), and
below range (<70 and <54 mg/dL). The number of
hyperglycemic excursions >180 mg/dL was also cal-
culated, where the first data point >180 mg/dL counts
but the rest do not count for that excursion.
Calculation of total daily dose of insulin, average car-
bohydrate intake, and average carbohydrate entries
per day.
Analysis of a variety of clinically approved GV met-
rics11-13 such as low blood glucose index (LBGI), high
blood glucose index (HBGI), coefficient of variation
(CV), standard deviation (SD), and J_index (which
stresses both the importance of the mean level and
variability of glycemic levels)14 before and after meals
for both pre- and post-PERT times, for one hour (0-60
minutes), two hours (0-120 minutes), and three hours
(0-180 minutes) each. The time slices before and after
meals were identified by selecting glucose entries
after all carbohydrate entries >15 g (therefore exclud-
ing hypoglycemia-related treatment <15 g). The time
slices were truncated when the onset of a high tempo-
rary target was detected, as that indicated exercise was
forthcoming, or if another carbohydrate entry >15 g
appeared during the time window. Each experiment
was repeated for one, two, and three-hour slices before
and after the meal.
Glucose outcomes and variability assessment (mean
of LBGI, HBGI, CV, SD, J_index, as well as mean
TIR/time out of range [TOR] metrics)15 both for all
meals overall, as well as breakfast-only (because it
was an identical meal across both time periods).
Analysis of post-PERT data including total daily fat
and protein intake, total daily PERT dosing, and addi-
tional lipase usage.
(Analysis scripts used for glucose outcomes and variability
are open source.)16
Results
A time in range exceeding recommended goals from ADA
2022 Standards of Care17 (>70%) was observed in this pre-
PERT data set, at 92.12% (70-180 mg/dL), and it further
improved to 93.70% following PERT commencement. Time
below 70 mg/dL decreased from 4.11% to 3.84%, and time
above 180 mg/dL similarly decreased, 3.77% to 2.46%. The
number of unique excursions >180 mg/dL dropped from 40
Lewis and Shahid 3
(pre-PERT) to 26 (post-PERT). Time below 54 mg/dL also
exceeded the recommended goal (<1%) but did not change
significantly between the two time periods (0.56% and
0.68%). These data were not further evaluated to exclude
potential “compression lows.”
There is a positive autocorrelation found among the glu-
cose data, which is an indication of the ability to conduct
detailed time series analysis. Figure A (A and B) and Figure
B (A and B) (Supplemental Appendix) demonstrate that pat-
terns for both ACF and PACF among glucose profiles pre-
and post-PERT remain the same.
Table 1 provides a comparison and summary of pre- and
post-PERT data of both glycemic and macronutrient-related
metrics. Total daily fat and protein intake were only recorded
after PERT commencement and averaged 115 g of fat and 65
g of protein daily. Figure C (Supplemental Appendix) illus-
trates the breakdown of average daily energy intake in the
post-PERT period, and Figure D (Supplemental Appendix)
illustrates the daily totals of each macronutrient by day over
the 28 days post-PERT period. The average daily carbohy-
drate intake slightly decreased from 183 g (pre-PERT) to 166
g (post-PERT), but continued returning to the previous aver-
age (Figure E, Supplemental Appendix directly compares the
daily carbohydrate intake across 28 days both pre- and post-
PERT). The average number of carbohydrate entries per day
(6.03-4.89) decreased, whereas the average grams of carbo-
hydrate per entry for entries >15 g (42-50) increased in the
post-PERT time period.
Figure F (Supplemental Appendix) plots the amount of
lipase taken per day, showing an increasing trend over time,
compared with the number of PERT and standalone lipase
pills consumed.
Table 2 provides all calculated GV-related variables.
Figure 1 visualizes mean LBGI, HBGI, TOR > 180, TOR <
70, and TOR < 54 for both pre- and post-PERT time periods,
for all meals and also for breakfast only. (All other data plots
are available in Figure G in Supplemental Appendix). In
Figure 2, the rate of change (RoC) in glucose (mg/dL/min-
ute) is plotted with a visible difference for the reduction in
glucose RoC per minute in post-PERT (ie, sigma = 0.91
post-PERT vs 1.03 pre-PERT), indicating less impact to glu-
cose levels in meals with PERT. Figure 3 plots glucose data
for the entire time periods of pre- and post-PERT with a vis-
ible reduction in the frequency of TOR > 180 mg/dL. The
total daily dose of insulin decreased slightly (31.8-29.0
units), matching the decrease in carbohydrate intake during
the commencement of PERT.
Discussion
This article provides an opportunity to analyze glycemic out-
comes and time-in-range improvements in an individual with
type 1 diabetes using an open-source AID system with newly
diagnosed EPI necessitating the commencement of PERT. The
available real-world data for n = 1 enables a demonstration of
a variety of methods of data analysis that could be used to
design further studies in the population of people with insulin-
requiring diabetes and EPI to address the knowledge gap that
exists with regard to glycemic changes with the onset of PERT.
Although this n = 1 individual already has achieved time
in range above recommended goals from ADA 2022
Standards of Care11 (>70% TIR, <4% <70 mg/dL, <1%
<54 mg/dL) before PERT, the commencement of PERT
slightly improved glycemic outcomes further. Average daily
carbohydrate intake was slightly reduced at the start of the
post-PERT time period, likely as a result of needing to dose
additional medication (pancrelipase) for all food consumed.
This is also illustrated by the change in the number of total
carbohydrate entries per day decreasing (6.03 pre-PERT;
4.89 post-PERT), but as seen in Figure E, carbohydrate
intake was trending upward toward the previous levels. The
average grams of carbohydrate per entry for entries >15 g
Table 1. Pre- and Post-PERT Metrics.
Pre-PERT Post-PERT
TIR 70-180 (%) 92.12 93.70
TOR > 180 (%) 3.77 2.46
TOR < 70 (%) 4.11 3.84
TOR < 54 (%) 0.56 0.68
TDD insulin (units) 31.8 29.0
Total average fat intake (g) Not recorded 115
Total average protein intake (g) Not recorded 65
Total average daily carbohydrate intake (g) 183 166
Average number of daily carbohydrate entries 6.03 4.89
Average number of daily carbohydrate corrections (<15 g) 1.85 1.71
Average number of meals (>15 g carbohydrate) 4.18 3.18
Average carbohydrate entry of meals (>15 g carbohydrate) 42 50
Average miles run per week 18.3 20.3
Abbreviations: PERT, pancreatic enzyme replacement therapy; TDD, total daily dose; TIR, time in range; TOR, time out of range.
4 Journal of Diabetes Science and Technology 00(0)
Table 2. Mean CV, SD, TOR (<54, <70, >180), TIR, J_Index, HBGI, and LBGI Before and After PERT.
Mean CV
(pre-PERT before
meal)
Mean CV
(pre-PERT after
meal)
Mean CV
(post-PERT before
meal)
Mean CV
(post-PERT after
meal)
Mean CV
(pre-PERT before
breakfast)
Mean CV
(pre-PERT after
breakfast)
Mean CV
(post-PERT before
breakfast)
Mean CV
(post-PERT after
breakfast)
0-1 h 6.49 5.84 5.78 5.78 7.92 5.79 6.08 6.85
0-2 h 9.27 8.38 8.06 9.04 10.70 9.51 8.61 9.21
0-3 h 11.46 10.51 10.39 11.48 12.75 12.74 11.19 12.57
Mean SD
(pre-PERT before
meal)
Mean SD
(pre-PERT after
meal)
Mean SD
(post-PERT before
meal)
Mean SD
(post-PERT after
meal)
Mean SD
(pre-PERT before
breakfast)
Mean SD
(pre-PERT after
breakfast)
Mean SD
(post-PERT before
breakfast)
Mean SD
(post-PERT after
breakfast)
0-1 h 7.15 6.25 6.11 6.05 7.88 6.02 5.82 6.38
0-2 h 10.20 8.96 8.65 9.44 10.63 10.08 8.51 8.85
0-3 h 12.91 11.24 11.18 12.02 12.83 13.82 11.19 12.56
Mean TOR <
70 [%]
(pre-PERT before
meal)
Mean TOR <
70 [%]
(pre-PERT after
meal)
Mean TOR <
70 [%]
(post-PERT before
meal)
Mean TOR <
70 [%]
(post-PERT after
meal)
Mean TOR < 70 [%]
(pre-PERT before
breakfast)
Mean TOR <
70 [%]
(pre-PERT after
breakfast)
Mean TOR < 70 [%]
(post-PERT before
breakfast)
Mean TOR <
70 [%]
(post-PERT after
breakfast)
0-1 h 2.87 1.80 5.10 6.37 9.88 5.25 10.12 13.99
0-2 h 3.50 1.23 4.71 5.44 9.41 3.09 9.97 11.01
0-3 h 3.18 2.04 3.92 5.10 7.30 5.86 7.54 8.73
Mean TOR>180
[%]
(pre-PERT before
meal)
Mean TOR>180
[%]
(pre-PERT after
meal)
Mean TOR>180
[%]
(post-PERT before
meal)
Mean TOR>180
[%]
(post-PERT after
meal)
Mean TOR>180 [%]
(pre-PERT before
breakfast)
Mean TOR>180
[%]
(pre-PERT after
breakfast)
Mean TOR>180 [%]
(post-PERT before
breakfast)
Mean TOR>180
[%]
(post-PERT after
breakfast)
0-1 h 1.00 1.40 1.18 0.49 1.54 5.25 3.27 0.00
0-2 h 1.03 1.53 1.18 0.54 0.77 5.25 3.42 0.74
0-3 h 1.27 1.20 1.05 1.14 0.51 3.91 3.08 1.69
Mean TIR [%]
(pre-PERT before
meal)
Mean TIR [%]
(pre-PERT after
meal)
Mean TIR [%]
(post-PERT before
meal)
Mean TIR [%]
(post-PERT after
meal)
Mean TIR [%]
(pre-PERT before
breakfast)
Mean TIR [%]
(pre-PERT after
breakfast)
Mean TIR [%]
(post-PERT before
breakfast)
Mean TIR [%]
(post-PERT after
breakfast)
0-1 h 96.13 96.80 93.73 93.14 88.58 89.51 86.61 86.01
0-2 h 95.47 97.23 94.12 94.02 89.81 91.67 86.61 88.24
0-3 h 95.56 96.76 95.03 93.76 92.18 90.23 89.38 89.58
Mean J_index
(pre-PERT before
meal)
Mean J_index
(pre-PERT after
meal)
Mean J_index
(post-PERT before
meal)
Mean J_index
(post-PERT after
meal)
Mean J_index
(pre-PERT before
breakfast)
Mean J_index
(pre-PERT after
breakfast)
Mean J_index
(post-PERT before
breakfast)
Mean J_index
(post-PERT after
breakfast)
0-1 h 13.73 13.02 12.88 12.80 13.29 13.52 11.33 10.76
0-2 h 14.76 13.72 13.77 13.46 13.60 14.84 12.54 11.80
0-3 h 15.75 14.02 14.54 14.13 13.70 15.68 13.42 13.42
Mean LBGI
(pre-PERT before
meal)
Mean LBGI
(pre-PERT after
meal)
Mean LBGI
(post-PERT before
meal)
Mean LBGI
(post-PERT after
meal)
Mean LBGI
(pre-PERT before
breakfast)
Mean LBGI
(pre-PERT after
breakfast)
Mean LBGI
(post-PERT before
breakfast)
Mean LBGI
(post-PERT after
breakfast)
0-1 h 1.32 1.31 1.87 1.70 2.80 2.36 3.35 3.14
0-2 h 1.30 1.17 1.62 1.61 2.56 1.75 2.78 2.66
0-3 h 1.24 1.27 1.44 1.59 2.24 1.93 2.33 2.32
Mean HBGI
(pre-PERT before
meal)
Mean HBGI
(pre-PERT after
meal)
Mean HBGI
(post-PERT before
meal)
Mean HBGI
(post-PERT after
meal)
Mean HBGI
(pre-PERT before
breakfast)
Mean HBGI
(pre-PERT after
breakfast)
Mean HBGI
(post-PERT before
breakfast)
Mean HBGI
(post-PERT after
breakfast)
0-1 h 0.46 0.48 0.49 0.43 0.62 1.14 0.60 0.25
0-2 h 0.53 0.47 0.50 0.41 0.46 1.09 0.61 0.36
0-3 h 0.63 0.42 0.50 0.46 0.34 0.98 0.57 0.53
Mean TOR <
54 [%]
(pre-PERT before
meal)
Mean TOR <
54 [%]
(pre-PERT after
meal)
Mean TOR <
54 [%]
(post-PERT before
meal)
Mean TOR <5
4 [%]
(post-PERT after
meal)
Mean TOR < 54 [%]
(pre-PERT before
breakfast)
Mean TOR< 54
[%]
(pre-PERT after
breakfast)
Mean TOR < 54 [%]
(post-PERT before
breakfast)
Mean TOR <
54 [%]
(post-PERT after
breakfast)
0-1 h 0.93 0.33 1.37 1.27 4.17 1.54 4.17 1.54
0-2 h 0.83 0.23 1.18 0.64 3.57 0.77 3.57 0.77
0-3 h 0.67 0.20 0.78 0.69 2.38 0.72 2.38 0.72
Abbreviations: CV, coefficient of variation; HBGI, high blood glucose index; LBGI, low blood glucose index; PERT, pancreatic enzyme replacement therapy; SD, standard
deviation; TIR, time in range; TOR, time out of range.
Lewis and Shahid 5
(42-50) increased in the post-PERT time period, a metric that
correlates with successfully using PERT to digest larger
quantities of food than previously. As seen in Figure D, the
intake of carbohydrates slightly increased throughout the
post-PERT time period, indicating that the reduction in car-
bohydrate intake was likely a short-term change influenced
Figure 1. Plots of mean LBGI, HBGI, TOR > 180, TOR < 70, TOR < 54 before and after all meals and before and after breakfast
only. Abbreviations: HBGI, high blood glucose index; LBGI, low blood glucose index; PERT, pancreatic enzyme replacement therapy;
TOR, time out of range.
6 Journal of Diabetes Science and Technology 00(0)
by the adjustment to taking PERT. Fat and protein consump-
tion trended slightly down throughout the post-PERT time
period (protein R2 = 0.11; fat R2 = 0.097, carbohydrate
R2 = 0.051).
Figure F (Supplemental Appendix) illustrates the relation-
ship between the total number of units of lipase per day and
the number of PERT-related pills consumed per day to the
total number of units of lipase. It is possible to have different
levels of pill burden depending on the choice for how to con-
sume lipase as standalone, over the counter (smaller) lipase-
only pills versus combination enzyme, prescription
pancrelipase pills. PERT is similar to insulin in that it requires
self-titration depending on the food consumed, so the quan-
tity needed also varies at each meal. Some guidelines recom-
mend starting at 50 000 lipase/meal, but symptom resolution
drives the titration strategy of PERT. Given that this indi-
vidual has moderate EPI and effective symptom reduction,
these doses, despite being lower than guidelines, are cur-
rently perceived as optimal for this individual.
GV is useful for assessing post-PERT changes alongside
symptom reduction, especially in an individual with already
above-goal TIR. We assessed GV metrics for meals overall
and specifically for breakfast only, since the meal was identi-
cal (estimated 54 g of fat, 16 g of protein, and around 50 g of
carbohydrates) for both pre- and post-PERT time periods,
providing a useful illustration of PERT-related impacts on
glucose.
Mean HBGI was slightly reduced after meals with PERT
compared with pre-PERT. However, there was a more notice-
able reduction from HBGI before (eg 1.14 in the first hour)
and after PERT commencement (0.25 in the first hour) in
breakfast alone (Figure 1). The changes in post-meal glucose
outcomes from the same meal with the same individual
and AID behavior indicate a positive impact from PERT
commencement. This PERT impact on breakfast is also illus-
trated by the mean TOR > 180 reductions after meals overall
(1.40% in the first hour following meals before PERT and
0.49% in the first hour following meals with PERT). PERT
essentially eliminates any first hour >180 excursions fol-
lowing breakfast (5.25% after breakfast without PERT to
0.0% in the first hour following breakfast with PERT). This
may illustrate how PERT helps the body more effectively
digest fat and protein, so carbohydrate absorption overall is
more mediated than it was before PERT commencement.
The overall glucose metrics indicated a slight overall
decrease in TOR < 70 and an even more slight increase in
TOR < 54. Most of the increase in TOR < 54 may be attrib-
uted to the post-breakfast time versus other meals in the day
(Figure 1). Breakfast was at a variable time each day, as early
as 4 am (an outlier) but more commonly between 8 and 11 am.
The post-breakfast window is also when the individual chose
to run five days of the week (an average of 18 or 20 miles per
week in each time period). Attempts were made to clean the
data and truncate the meal observation period any time a
temporary target was enacted before/during exercise, but that
temporary target entry was not consistently applied and
therefore some of these data may be attributable to exercise-
induced hypoglycemia rather than indicating that PERT was
correlated with inducing more hypoglycemia.
The overall interday glucose SD dropped from 30.68 to
29.28 mg/dL and CV dropped from 27.30 to 25.92. The mean
SD and CV are lowest for one hour as compared with two
and three hours following the meals with and without PERT,
which reflects the fact day-to-day variation in postprandial
blood sugar is larger than variability of blood sugar at meal-
time. Without PERT, the SD is reduced with the overall meal
intake for all periods. The pattern (decrease in SD) follows in
the case of breakfasts for the first one and two hours follow-
ing. However, there is a slight increase in SD after meals
with PERT for the three hours following. Similarly, the CV
without PERT is reduced with the meal intake but slightly
increases in the case of PERT for each expanded time slice.
The autocorrelation (see Figure A, both A and B, in
Supplemental Appendix) become statistically insignificant
around 60 to 70 glucose data points, which may be illustra-
tive of the duration of insulin activity (DIA) being more than
five hours for this individual, with each data point represent-
ing five minutes.
The changes in GV and achieving high TIR without sig-
nificant hypoglycemia provide another indicator of PERT
success. PERT is typically assessed as successful based on
reduction in malabsorption over time, which can be hard to
measure, and symptom resolution as self-reported by
patients. Thus, GV analysis in people with diabetes can be a
useful and possible shorter-term indicator of PERT efficacy.
It could also be useful for assessing the presence of EPI in
someone with suspected insufficiency who commences
PERT before elastase testing. Meanwhile, because it is com-
mon for PERT to be underdosed and because it can be hard
Figure 2. Illustration of glucose rate of change differences
before and after PERT commencement, indicating less impact to
glucose levels in meals with PERT. Abbreviation: PERT, pancreatic
enzyme replacement therapy.
Lewis and Shahid 7
to self-titrate, GV changes could be an earlier indicator of
PERT efficacy and serve as encouragement for individuals
with EPI to continue PERT titration and arrive at optimal
dosing.
In this n = 1 example, the individual recognized and
sought a diagnosis for EPI and treatment with PERT, but it
took two years to identify EPI as the cause for new-onset
gastrointestinal symptoms. There are high rates of EPI in dia-
betes, and many people go undiagnosed for long periods, as
is true with other common gastrointestinal conditions that
are frequently present in people with diabetes. This is yet
another indicator that glycemic outcomes may be influenced
by variables outside of an individual’s control5 and ability to
self-manage (eg someone with EPI can’t make more
enzymes), and points to the need for further study of many of
these GI-related conditions and their impact on glycemic
outcomes and GV. The use of AID and CGM should be
funded and encouraged for studies in such related gastroin-
testinal conditions to fill the knowledge gaps about glycemic
impacts of medications such as those used in PERT.
Previous studies found that open-source AID effectively
adjusts for a lot of “noise,”18 and this n = 1 study adds the
finding that it also responds effectively to variable digestion
due to pancreatic insufficiency. This study demonstrates a
viable method for improving the understanding of glycemic
response to medications that influence digestion, such as
PERT and many other medications commonly used by peo-
ple with diabetes. Further studies should address glycemic
change in response to PERT in a broader population of peo-
ple living with insulin-requiring diabetes, and further stud-
ies should be done on other types of commonly used
medications by people with insulin-requiring diabetes to
improve the understanding of the glycemic impact of these
medications.
Limitations
There are a few limitations specific to the design of this data
analysis. Most importantly, this is an n = 1 retrospective
analysis rather than a clinical study, and therefore defini-
tively not meant to be representative of all PwD, people
using AID, PwD with EPI, or people with EPI in general.
Instead, it should be used as an example of the type of analy-
ses that are possible, and to inform study design for future
studies with larger populations. As an n = 1 study, it was also
infeasible to estimate statistical significance of changes
observed, which would require larger or longer studies.
Figure 3. Plot of all glucose in each pre- and post-PERT time period, illustrating reduction in number of unique TOR > 180 excursions
from 40 to 26 following PERT commencement. Abbreviations: PERT, pancreatic enzyme replacement therapy; TOR, time out of range.
8 Journal of Diabetes Science and Technology 00(0)
The n = 1 example is also in someone achieving above-
goal glycemic outcomes and time in range, so absolute glyce-
mic and TIR changes with PERT are somewhat small, and also
likely influenced by the level of severity of EPI, which is con-
sidered to be “moderate” in this individual. For someone with
severe EPI, categorized by lower levels of elastase, different
TIR and glycemic outcomes prior to PERT onset, or different
levels of exercise or physical activity, the changes might be
more distinct. In addition, the timing of regular running five
days a week in the two to three hours following the breakfast
meal likely influenced the after-breakfast GV metrics both
pre- and post-PERT for this individual. Total daily fat and pro-
tein intake were only recorded after PERT commencement (as
well as carbohydrate intake), whereas carbohydrate intake is
the only dietary data available in the pre-PERT period.
Finally, this is not meant to purport to represent what should
be done nutrition-wise, energy-wise, or PERT-wise, but solely
as an example method of analysis for examining real-world
behavior and glycemic-related changes in someone with type
1 diabetes following the initial commencement of PERT and
throughout the first four weeks of PERT titration.
Conclusion
It is possible to use GV to assess changes in glycemic out-
comes in response to new-onset medications, such as PERT
in people with EPI and insulin-requiring diabetes. More
studies should use AID and CGM data to assess changes in
glycemic outcomes and variability to add to the knowledge
base of how medications affect glucose levels for people
with diabetes. Specifically, this n = 1 data analysis demon-
strates that GV can be useful for assessing the response to
PERT in someone with suspected or newly diagnosed EPI
and provide additional data points regarding the efficacy of
PERT titration over time.
Abbreviations
AID, automated insulin delivery; EPI or PEI, exocrine pancreatic
insufficiency; GV, glycemic variability; PERT, pancreatic enzyme
replacement therapy; TIR, time in range; TOR, time out of range.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect
to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, author-
ship, and/or publication of this article.
ORCID iDs
Dana M. Lewis https://orcid.org/0000-0001-9176-6308
Arsalan Shahid https://orcid.org/0000-0002-3748-6361
Supplemental Material
Supplemental material for this article is available online.
References
1. Mohapatra S, Majumder S, Smyrk TC, et al. Diabetes mellitus
is associated with an exocrine pancreatopathy: conclusions
from a review of literature. Pancreas. 2016;45(8):1104-
1110.
2. Duggan SN, Smyth ND, Murphy A, Macnaughton D, O’Keefe
SJ, Conlon KC. High prevalence of osteoporosis in patients
with chronic pancreatitis: a systematic review and meta-analy-
sis. Clin Gastroenterol Hepatol. 2014;12(2):219-228.
3. Domínguez-Muñoz JE. Pancreatic enzyme therapy for pan-
creatic exocrine insufficiency. Curr Gastroenterol Rep.
2007;9(2):116-122.
4. Radlinger B, Ramoser G, Kaser S. Exocrine pancreatic
insufficiency in type 1 and type 2 diabetes. Curr Diab Rep.
2020;20(6):18.
5. Lewis D. Quantifying input behaviors that influence clinical
outcomes in diabetes and other chronic illnesses. J Diabetes Sci
Technol. 2022;16(3):786-787.
6. Lewis D. History and perspective on DIY closed looping. J
Diabetes Sci Technol. 2019;13(4):790-793.
7. Lee JM, Newman MW, Gebremariam A, et al. Real-
world use and self-reported health outcomes of a patient-
designed do-it-yourself mobile technology system for
diabetes: lessons for mobile health. Diabetes Technol Ther.
2017;19(4):209-219.
8. Nightscout Data Transfer Tool [Internet]. https://github.
com/NightscoutFoundation/dataxfer. Cited March 31, 2022.
Accessed June 18, 2022.
9. Greshake Tzovaras B, Angrist M, Arvai K, et al. Open Humans:
a platform for participant-centered research and personal data
exploration. Gigascience. 2019;8(6):giz076.
10. Lewis D. Open Humans Data Tools: Tools to work with data
downloaded from Open Humans research platform [Internet].
https://github.com/danamlewis/OpenHumansDataTools/.
Cited March 31, 2022. Accessed June 18, 2022.
11. Kovatchev B. Glycemic variability: risk factors, assessment,
and control. J Diabetes Sci Technol. 2019;13(4):627-635.
12. Zhou Z, Sun B, Huang S, Zhu C, Bian M. Glycemic variability:
adverse clinical outcomes and how to improve it? Cardiovasc
Diabetol. 2020;19(1):102.
13. Rodbard D. Glucose variability: a review of clinical applica-
tions and research developments. Diabetes Technol Ther.
2018;20(S2):S25-S215.
14. Wójcicki JM. “J”-index. A new proposition of the assessment
of current glucose control in diabetic patients. Horm Metab
Res. 1995;27(1):41-42.
15. Shahid A, Lewis DM. Large-scale data analysis for glucose
variability outcomes with open-source automated insulin deliv-
ery systems. Nutrients. 2022;14(9):1906.
16. Shahid A, Lewis D. GV-pre-post-analysis Scripts in
OpenHumansDataTools [Internet]. https://github.com/danam-
lewis/OpenHumansDataTools/tree/master/bin/GV-pre-post-
analysis. Cited April 6, 2022. Accessed June 18, 2022.
17. American Diabetes Association Professional Practice
Committee, Draznin B, Aroda VR, et al. 6. Glycemic targets:
standards of medical care in diabetes—2022. Diabetes Care.
2022;45(suppl 1):S83-S96.
18. Lewis DM. Do-it-yourself artificial pancreas system and the
OpenAPS movement. Endocrinol Metab Clin North Am.
2020;49(1):203-213.
ResearchGate has not been able to resolve any citations for this publication.
Article
Full-text available
Open-source automated insulin delivery (AID) technologies use the latest continuous glucose monitors (CGM), insulin pumps, and algorithms to automate insulin delivery for effective diabetes management. Early community-wide adoption of open-source AID, such as OpenAPS, has motivated clinical and research communities to understand and evaluate glucose-related outcomes of such user-driven innovation. Initial OpenAPS studies include retrospective studies assessing high-level outcomes of average glucose levels and HbA1c, without in-depth analysis of glucose variability (GV). The OpenAPS Data Commons dataset, donated to by open-source AID users with insulin-requiring diabetes, is the largest freely available diabetes-related dataset with over 46,070 days’ worth of data and over 10 million CGM data points, alongside insulin dosing and algorithmic decision data. This paper first reviews the development toward the latest open-source AID and the performance of clinically approved GV metrics. We evaluate the GV outcomes using large-scale data analytics for the n = 122 version of the OpenAPS Data Commons. We describe the data cleaning processes, methods for measuring GV, and the results of data analysis based on individual self-reported demographics. Furthermore, we highlight the lessons learned from the GV outcomes and the analysis of a rich and complex diabetes dataset and additional research questions that emerged from this work to guide future research. This paper affirms previous studies’ findings of the efficacy of open-source AID.
Article
Full-text available
Abstract Glycemic variability (GV), defined as an integral component of glucose homoeostasis, is emerging as an important metric to consider when assessing glycemic control in clinical practice. Although it remains yet no consensus, accumulating evidence has suggested that GV, representing either short-term (with-day and between-day variability) or long-term GV, was associated with an increased risk of diabetic macrovascular and microvascular complications, hypoglycemia, mortality rates and other adverse clinical outcomes. In this review, we summarize the adverse clinical outcomes of GV and discuss the beneficial measures, including continuous glucose monitoring, drugs, dietary interventions and exercise training, to improve it, aiming at better addressing the challenging aspect of blood glucose management.
Article
Full-text available
Purpose of review: Type 1 and type 2 diabetes are often accompanied by mostly mild forms of exocrine pancreatic insufficiency. Despite high prevalence, little is known about the clinical consequences of exocrine pancreatic insufficiency and its optimal (nutritional) treatment. Even less is known if and to what extent exocrine pancreas insufficiency also affects glycemic control in diabetes. This article aims for summarizing current clinical knowledge on screening, diagnosis, and treatment and gives an overview on the pathophysiology of exocrine pancreatic insufficiency in diabetes. Recent findings: Recent studies reveal novel insights into the close interaction of acinar, ductal, and endocrine cells and the gut-pancreas axis. Exocrine pancreatic insufficiency is a clinically relevant, frequent but poorly understood disorder in both type 1 and type 2 diabetes.
Article
Full-text available
Background: Many aspects of our lives are now digitized and connected to the internet. As a result, individuals are now creating and collecting more personal data than ever before. This offers an unprecedented chance for human-participant research ranging from the social sciences to precision medicine. With this potential wealth of data comes practical problems (e.g., how to merge data streams from various sources), as well as ethical problems (e.g., how best to balance risks and benefits when enabling personal data sharing by individuals). Results: To begin to address these problems in real time, we present Open Humans, a community-based platform that enables personal data collections across data streams, giving individuals more personal data access and control of sharing authorizations, and enabling academic research as well as patient-led projects. We showcase data streams that Open Humans combines (e.g., personal genetic data, wearable activity monitors, GPS location records, and continuous glucose monitor data), along with use cases of how the data facilitate various projects. Conclusions: Open Humans highlights how a community-centric ecosystem can be used to aggregate personal data from various sources, as well as how these data can be used by academic and citizen scientists through practical, iterative approaches to sharing that strive to balance considerations with participant autonomy, inclusion, and privacy.
Article
Full-text available
People with diabetes have been experimenting and self-modifying diabetes devices and technologies for many decades, in order to achieve the best possible quality of life and improving their long-term outcomes. There are now hundreds of individuals using DIY closed loop systems globally. They work similarly to commercial systems at a basic level, automatically adjusting and controlling insulin dosing, but are different in terms of transparency, access, customization, and usability. The potential downsides to DIY closed looping include varying responses from individual HCPs, who may be concerned about their own liability. However, initial outcomes from this self-selected community (including adult and pediatric populations globally) have been positive. There have now been several studies documenting improvements in A1c, time in range, and other outcomes such as quality-of-life benefits. More studies on quality-of-life improvements and more collaboration between companies and the community are recommended.
Article
The American Diabetes Association (ADA) "Standards of Medical Care in Diabetes" includes the ADA's current clinical practice recommendations and is intended to provide the components of diabetes care, general treatment goals and guidelines, and tools to evaluate quality of care. Members of the ADA Professional Practice Committee, a multidisciplinary expert committee (https://doi.org/10.2337/dc22-SPPC), are responsible for updating the Standards of Care annually, or more frequently as warranted. For a detailed description of ADA standards, statements, and reports, as well as the evidence-grading system for ADA's clinical practice recommendations, please refer to the Standards of Care Introduction (https://doi.org/10.2337/dc22-SINT). Readers who wish to comment on the Standards of Care are invited to do so at professional.diabetes.org/SOC.
Article
People with diabetes have been experimenting with and modifying their own diabetes devices and technologies for many decades in order to achieve the best possible quality of life and improving their long-term outcomes, including do-it-yourself (DIY) closed loop systems. Thousands of individuals use DIY closed loop systems globally, which work similarly to commercial systems by automatically adjusting and controlling insulin dosing, but are different in terms of transparency, access, customization, and usability. Initial outcomes seen by the DIY artificial pancreas system community are positive, and randomized controlled trials are forthcoming on various elements of DIYAPS technology.
Article
Glycemic variability (GV) a well-established risk factor for hypoglycemia and a suspected risk factor for diabetes complications. GV is also a marker of the instability of a person’s metabolic system, expressed as frequent high and low glucose excursions and overall volatile glycemic control. In this review, the author discusses topics related to the assessment, quantification, and optimal control of diabetes, including (1) the notion that optimal control of diabetes, that is, lowering of HbA1c—the commonly accepted gold-standard outcome—can be achieved only if accompanied by simultaneous reduction of GV; (2) assessment and visualization of the two principal dimensions of GV, amplitude and time, which is now possible via continuous glucose monitoring (CGM) and various metrics quantifying GV and the risks associated with hypo- and hyperglycemic excursions; and (3) the evolution of diabetes science and technology beyond quantifying GV and into the realm of GV control via pharmacological agents, for example, GLP-1 receptor agonists and DPP-4 inhibitors, which have pronounced variability-reducing effect, or real-time automated closed-loop systems commonly referred to as the “artificial pancreas.” The author concludes that CGM allows close tracking over time, and therefore precise quantification, of glycemic variability in diabetes. The next step—optimal control of glucose fluctuations—is also taken by medications with pronounced GV-lowering effect primarily in type 2 diabetes, and by automated insulin delivery in type 1 diabetes. Contemporary CGM-based artificial pancreas systems use specific GV representations as input signals, and thus their main objective is to minimize GV and, from there, optimize glycemic control.
Article
Glycemic variability (GV) is a major consideration when evaluating quality of glycemic control. GV increases progressively from prediabetes through advanced T2D and is still higher in T1D. GV is correlated with risk of hypoglycemia. The most popular metrics for GV are the %Coefficient of Variation (%CV) and standard deviation (SD). The %CV is correlated with risk of hypoglycemia. Graphical display of glucose by date, time of day, and day of the week, and display of simplified glucose distributions showing % of time in several ranges, provide clinically useful indicators of GV. SD is highly correlated with most other measures of GV, including interquartile range, mean amplitude of glycemic excursion, mean of daily differences, and average daily risk range. Some metrics are sensitive to the frequency, periodicity, and complexity of glycemic fluctuations, including Fourier analysis, periodograms, frequency spectrum, multiscale entropy (MSE), and Glucose Variability Percentage (GVP). Fourier analysis indicates progressive changes from normal subjects to children and adults with T1D, and from prediabetes to T2D. The GVP identifies novel characteristics for children, adolescents, and adults with type 1 diabetes and for adults with type 2. GVP also demonstrated small rapid glycemic fluctuations in people with T1D when using a dual-hormone closed-loop control. MSE demonstrated systematic changes from normal subjects to people with T2D at various stages of duration, intensity of therapy, and quality of glycemic control. We describe new metrics to characterize postprandial excursions, day-to-day stability of glucose patterns, and systematic changes of patterns by day of the week. Metrics for GV should be interpreted in terms of percentiles and z-scores relative to identified reference populations. There is a need for large accessible databases for reference populations to provide a basis for automated interpretation of GV and other features of continuous glucose monitoring records.