DTU Health Tech

Department of Health Technology

Oral insulin
Preclinical studies of the oral bioavailability and
toxicology of peptide­based drugs
22237 Preclinical Drug Development
Authors: Xiangyou Meng (s183463), Karoline Skree­Lassen (s184238), Joan Cortes Gimeno
(s212509), Ana Aguilar Anaya (s221294), Christoforos Anagnostopoulos (s222978),
Despoina Paloglou (s223180)

Date: 5th of December, 2022

Abstract
Diabetes is a chronic, metabolic disease whose incidence is rapidly increasing worldwide. In ad­
dition to being a great health burden for the afflicted, it is a major financial concern for society as
well. As of now, the primary treatment option is subcutaneous administration of insulin. Though
effective, this administration route is associated with a multitude of shortcomings, such as pain and
needle phobia, thus resulting in poor patient compliance.
This report investigates the potential preclinical tests that can be performed for a novel orally ad­
ministered insulin formulation to pass on to clinical trials. This study suggests conducting an in
silico study using the ACAT model to obtain an initial estimation of the dissolution and permeation
of the novel formulation. Subsequently, the study proposes further investigating the dissolution of
the formulation by employing a two­step dissolution study in combination with the TIM­1 model
to obtain more comprehensive information on the transit of the drug. These findings are further
supplemented with in vitro pH studies due to the decisive influence pH has on insulin as a com­
pound. Concurrently, it is suggested to examine the permeation of the drug by performing an in
vitro assay using Caco­2 cells. Following this, the study suggests performing in vivo tests in rats
to find the bioavailability of the novel formulation. Finally, both in vitro and in vivo toxicology
studies are proposed to investigate acute and chronic toxicity. All tests are carefully selected to
combined yield all the necessary information on the novel oral insulin formulation in order to pass
on to clinical trials.

ii Group 1: Oral insulin

Abbreviations
ACAT: Advanced Compartmental Absorption and Transit
ADME: Absorption, Distribution, Metabolism, Excretion
API: Active Pharmaceutical Ingredient
Caco­2: Cancer coli­2
CAT: Compartmental Absorption and Transit
cFaSSGF: Canine Fasted­State Simulated Gastric Fluid
cFaSSIF: Canine Fasted­State Simulated Intestinal Fluid
CYP: Cytochromes P450
CCK: Cell Counting Kit
CD: Circular Dichroism
ELISA: Enzyme­linked Immunosorbent Assay
EMA: European Medicines Agency
FDA: Food and Drug Administration
FITC: Fluorescein Isothiocyanate
GI: Gastrointestinal
GIT: Gastrointestinal Tract
GLP­1: Glucagon Like Peptide 1
IOAC: Intestine­On­A­Chip
IVIVC: In Vivo­In Vitro Correlation
NOAEL: No­Observed­Adverse­Effect Level
OECD: Organisation for Economic Cooperation and Development
P­gp: P­glycoprotein
pI: Isoelectric Point
PK: Pharmacokinetic
PKPD: Pharmacokinetics and Pharmacodynamics
PBPK: Physiologically­Based Pharmacokinetics
rFaSSGF: Rat Fasted­State Simulated Gastric Fluid
rFaSSIF: Rat Fasted­State Simulated Intestinal Fluid
TEER: Transepithelial Electrical Resistance
TIM: TNO Intestinal Model

Group 1: Oral insulin iii

Contents
Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ii
Abbreviations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iii

1 Introduction 2
1.1 Diabetes: A 21st century epidemic . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.2 Oral bioavailability of insulin: Physiological challenges . . . . . . . . . . . . . . 3

2 In silico models 4
2.1 Prediction of absorption and oral bioavailability using physiologically­based phar­
macokinetic modeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2.1.1 Physiologically­based pharmacokinetic models for predicting the absorp­
tion and oral bioavailability: CAT and ACAT models . . . . . . . . . . . 4

3 In vitro tests 6
3.1 Permeability testing using Caco­2 cell monolayers . . . . . . . . . . . . . . . . . 6
3.1.1 Rapid 3­day system for Caco­2 monolayers: FITC­insulin permeation . . 7
3.2 In vitro dissolution models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
3.2.1 Dissolution testing using a two­step dissolution model . . . . . . . . . . 8
3.2.2 Dissolution testing using a TIM­1 model . . . . . . . . . . . . . . . . . 9
3.3 In vitro pH studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
3.4 In vitro toxicity testing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
3.4.1 Cytotoxicity testing in Caco­2 cells . . . . . . . . . . . . . . . . . . . . 10

4 In vivo assays 11
4.1 In vivo pharmacology studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
4.2 In vivo toxicity studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

5 Discussion of selected methods 13

Bibliography 16

Group 1: Oral insulin 1

1 Introduction
1.1 Diabetes: A 21st century epidemic
Diabetes is a chronic disease that has had a rapid increase in incidents throughout the last decades.
From 2019 to 2021, the number of people with diabetes increased by a striking 16%, thus currently
affecting one in ten adults worldwide [1]. The rise in incidents does not only carry a great health
and financial burden for the patients but society as a total [2].
Diabetes is a metabolic disease that is characterized by hyperglycemia as a result of complete or
relative failure to secrete insulin [2]. Insulin is the hormone that regulates human metabolism by
promoting the uptake of glucose in the liver, muscles, and fat [3]. The two most common types of
diabetes are type 1 and type 2 diabetes: Type 1 diabetes is mostly caused by genetics since it arises
when the pancreas ceases to produce insulin because the beta cells responsible for the production
are damaged by an autoimmune mechanism. The condition ensues over a period of several years,
usually during childhood or youth [2, 4], and requires continuous treatment with insulin to prevent
acute complications [5]. Type 2 diabetes is by far the most abundant type of diabetes, accounting
for more than 90% of incidents. It develops when the metabolic system becomes resistant to the
release of insulin. There are various degrees of insulin resistance [2], but the condition often
worsens throughout the patient’s lifetime [4]. Thus, the treatment regimen for patients with type 2
diabetes changes throughout the span of the disease. In the early stage of the disease, oral agents
are sufficient to achieve glycemic control, but subcutaneous treatment with insulin­analouges or
GLP­1­analouges will be inevitable at later stages [3].
Although subcutaneous administration has proven to be the most effective, this route is associated
with a multitude of side effects, such as pain, needle phobia, and, perhaps most crucially, poor
patient compliance [6, 5]. A study by Garcia­Perez et al. discovered that adherence to therapy
among patients with type 2 diabetes is achieved by less than 50% [7]. Consequently, the risk of
long­term complications increases, which in turn amplifies the associated morbidity rates and re­
sults in substantial healthcare­associated costs [7]. One way to increase adherence among patients
would be to change the route of administration to oral, which is known for having the highest pa­
tient compliance. Albeit not a new idea, as it was proposed in early scientific papers from the 1920s
[8], there is still no option of orally administered insulin on the market due to the many obstacles
related to the absorption of oral insulin.
This report investigates a novel insulin formulation that is seemingly able to be administered suc­
cessfully as an oral tablet. Furthermore, the formulation seems to protect the insulin through the
stomach without degradation. Since insulin is not a novel biological entity, certain regulatory as­
pects do not need to be considered, such as efficacy and cross­reactions. Hence, the main goals
of the preclinical testing will focus on determining the bioavailability of the novel formulation to
establish a dosage. Additionally, due to the physiochemical properties of insulin, it will be rele­
vant to conduct permeation and dissolution studies. Lastly, due to the change in formulation and
dosage, it will be vital to perform toxicity studies as well to minimize the dose­related risks.
In order to perform a thorough analysis of the properties of this, this report proposes using a holistic
approach, combining a number of techniques and assays. Together, they cover both in silico, in
vitro, and in vivo­based methods, respectively.

2 Group 1: Oral insulin

1.2 Oral bioavailability of insulin: Physiological challenges
Like the majority of drugs, insulin must reach the blood circulation in order to have a therapeutic
effect. For orally administered drugs, there are multiple obstacles to overcome prior to this. In the
context of peptide­based drugs, such as insulin, these obstacles become even more complex. This
is mainly due to the nature of insulin, as they are large molecules with hydrophilic characteristics
[9].
The first great obstacle is the acidic environment in the stomach, which puts the peptide­based
drug at risk of denaturation. Even if the drug manages to bypass the gastric acids intact, there
is an array of other obstacles awaiting in the gastrointestinal tract (GIT). A major hindrance that
limits the bioavailability of peptides is the presystemic enzymatic degradation in the GIT [10].
This is caused by non­specific peptidases and enterocytes which effectively denature peptides.
Thus, only approximately 1% of administered peptides remain unscathed for absorption [11, 12].
Another major obstacle is the epithelial layer in the small intestine. The size and hydrophilicity
of peptide­like drugs make it very difficult for them to passively diffuse across the membrane by
either the transcellular or paracellular pathway [13]. Moreover, the epithelial layer lining the GIT
consists of columnar cells which are connected by tight junctions. Together with a layer of mucin,
this makes up a physical barrier that ensures minimal leakage across the membrane [11], as seen in
Fig. 1.1. Thusly, in order to cross the tight epithelium, peptide­based drugs must be broken down
to amino acids by gut enzymes whilst still retaining their biological activity.

Figure 1.1: Diagram illustrating the two major transport pathways of peptide­based compounds across the intestinal
epithelial membrane. The epithelial layer lining the GIT is a tightly bound collection of cells that ensure minimal
leakage. Columnar cells are tightly bound by hydrophobic proteins called occludes, which help make up this physical
barrier that must be overcome prior to absorption [11]. Additionally, a layer of mucin acts as a gel filtration membrane
that prevents the absorption of large molecules. The structure is further ensured by tight junctions that inhibit paracellular
transport. Illustration source: [10].

In addition to the aforementioned difficulties, one must also consider the naturally­occurring ef­
flux transporters in the intestine, such as P­glycoprotein (P­gp) [14]. These reside in the apical
membranes, which drugs must cross during transcellular absorption from the intestinal lumen to
the blood. Transporters partly function as a defense mechanism by limiting the absorption of for­
eign and potentially toxic compounds into the blood, simply by driving proteins that do manage to
pass into the cells back into the lumen [15, 16].
Finally, if the peptide­like drug succeeds at entering the systemic circulation, the presystemic
metabolism in the liver, more commonly known as the first pass effect, drastically decreases the
circulation time of the drug. Since these physiological challenges and parameters are increasingly
well­characterised, these data can be used favourably as input in in silico studies.

Group 1: Oral insulin 3

2 In silico models
2.1 Prediction of absorption and oral bioavailability using
physiologically­based pharmacokinetic modeling
Throughout the past decades, computational methods have become increasingly recognized for
their prediction of both the absorption and oral bioavailability of drugs. The models serve as tools
to integrate compound data from in vitro studies during the process of drug screening to provide
in vivo context [17]. Thusly, the application of pharmacokinetic (PK) modeling can be performed
advantageously during preclinical studies, even before the synthesis of the compound in question
[18, 19]. Applying predictive PK modeling during the early stages can be very beneficial, as one
can investigate the potential of the oral drug prior to conducting in vivo tests. Hence, the result
of the modeling functions as an early decision point for whether or not to continue studies on
the drug/new formulation based on the drug’s absorption profile. This is highly desirable since
low and highly variable intestinal absorption acts as a major cause of failure later in the pipeline
for alternative drug administration routes [20]. Yet another advantage of using PK modeling is
the surge in commercially available, user­friendly physiologically­based pharmacokinetic (PBPK)
model software, such as GastroPlus and SimCYP, which have become increasingly detailed [21].
The software can include a multitude of physiological parameters, making it possible to obtain
more accurate predictions. Additionally, computational modeling also acts as a promising alterna­
tive to more traditional experimental protocols for measuring and calculating the bioavailability,
which is much more costly and time­consuming in comparison ­ thus, many resources can be saved
by implementing in silico studies.
2.1.1 Physiologically­based pharmacokinetic models for predicting the absorption
and oral bioavailability: CAT and ACAT models
Among the vast options of computational models, PBPK models are especially promising for more
accurate prediction of absorption and oral bioavailability. Compared to certain other alternatives,
PBPK models are highly dynamic in nature and defined by a series of differential equations [22].
Various PK models with different strengths have been developed due to the complexity of oral
bioavailability, which is influenced by different inter­individual physiological and metabolic fac­
tors [22, 23]. One of particular interest is the compartment transit models, in which the intestinal
tract is modeled by a series of compartments [24]. For investigating insulin, two models are pro­
posed.
The Compartmental model of Absorption and Transit (CAT) model is a mechanistic model that
predicts the fraction of the absorbed dose (F a) of different drugs. The model mimics the intestinal
tract through a series of compartments [22]. In total, seven compartments together simulate the
various anatomical sections of the small intestine, cf. Fig. 2.1. The model considers the transit
time through the entire small intestine, which one can adjust by varying the transit rate constants,
kt [22]:

Figure 2.1: The compartmental absorption and transit model (CAT) accurately mimics the absorption and transit of drugs
through the small intestine. It consists of seven well­stirred compartments that combined mimic the gastro intestical
tract from the duodenum to the ileum. Compartment 1: Duodenum. Compartment 2­3: Jejenum. Compartment 4­7:
Ileum. Scheme source: [22]

4 Group 1: Oral insulin

In regard to low­permeability or low­solubility drugs, such as insulin, the predictions using CAT
can however be compromised [25]. One way to circumvent this is by using the Advanced Com­
partment Absorption and Transit (ACAT) model, which is an extended version of CAT, cf. Fig.
2.2. The ACAT model has two additional compartments, representing the stomach and colon seg­
ments as additional absorbing compartments, thus adding to the precision of the already existing
CAT model [22]. By doing so, it allows one to evaluate even more GIT processes, such as gastric
emptying and colonic absorption. Moreover, it has been proven useful to evaluate the effects of
changing pH­values by simply employing the widely accessible GastroPlus software [26]. Hence,
the usage of ACAT can allow for even more thorough explorations and more accurate predictions
for absorption and oral bioavailability.
Despite many advancements, accurate PKPD modeling still remains a challenge and requires re­
finement, as it is a relatively new approach [19]. Moreover, pharmacokinetic modeling is currently
not a requisite by regulatory authorities, even though the approach has great potential in saving both
time and money by virtue of its predictive capacities of relevant parameters for drug absorption.
Nevertheless, additional preclinical tests for investigating these parameters are vital and thus fur­
ther elaborated upon in the forthcoming sections.

Figure 2.2: Diagram depicting the Advanced Compartment Absorption and Transit (ACAT) model. ACAT is an ex­
tended version of the former Compartment Absorption and Transit Model, CAT, as it includes additional compartments,
making it possible to simulate stomach and colon absorption. In addition, it considers a wider range of metabolic phe­
nomena, such as drug dissolution, cell permeability, and first­pass metabolism both from the liver and the stomach. The
latter are depicted by the arrows denoted ”Clearance”. Diagram source: [22].

Group 1: Oral insulin 5

3 In vitro tests
3.1 Permeability testing using Caco­2 cell monolayers
A crucial step during preclinical testing of a novel oral insulin formulation would be to perform
permeability studies to properly assess the drug permeability. For this, Caco­2 cell cultures offer
a highly versatile in vitro method with great investigative usage in intestinal absorption studies.
This is important, as the majority of absorption takes place in the small intestines, partly due to its
large surface area and the presence of villi and microvilli [27].
The model consists of an apical chamber, an epithelial barrier, and a basolateral chamber. The api­
cal and basolateral chambers function as the donor and receiver chambers, respectively, cf. Fig.
3.1 [28]. The Caco­2 cell lines stem from colorectal adenocarcinoma (i.e. human colon cancer)
cells. When cultured, they spontaneously differentiate into enterocytes, thus forming a functional,
heterogeneous monolayer of cells that highly mimics the human intestinal epithelium [29, 30]. In
addition to possessing tight cellular junctions and microvilli, the monolayer expresses many of
the drug­metabolising enzymes involved in absorption processes, such as peptidases [31]. Fur­
thermore, the monolayer also possesses P­glycoprotein efflux transporters [31], which adds to the
model’s striking resemblance to actual physiological conditions. This makes them exceptionally
useful as permeability models.

Figure 3.1: A scheme depicting a Caco­2 monolayer in a transwell. The apical chamber functions as the donor com­
partment, whilst the basolateral chamber acts as the receiver compartment. One can quantify the transport across the
Caco­2 cell monolayer as an indicator of intestinal mucosa permeability. Illustration produced using BioRender.com.

Over the years, Caco­2 monolayers have become a standard screening method for the assessment
of intestinal drug permeability and absorption. Their usefulness in predicting the transport of drugs
via. the para­ and transcellular pathways have been repeatedly reported in the literature ­ also for
studying the permeability of drugs similar to oral insulin, i.e. drugs of BCS class III, which are
characterized by having poor lipophilicity/presumed high solubility and permeability, in addition
to undergoing P­glycoprotein efflux [31, 28].
However, there are certain limitations when using conventional Caco­2 cell assays. They are quite
time­consuming, as the cell line has to be cultivated for two­three weeks on semipermeable mi­
croporous filters on transwell plates. Moreover, this requires 9­10 labour­intensive cell feedings
[32, 6]. Another limitation of Caco­2 cells is that the tightness of the junctions is hypothesized
to surpass that of physiological endothelial cells. As a result, they exhibit too high transepithelial
electrical resistance (TEER)­values compared to in vivo values [6, 33].

6 Group 1: Oral insulin

3.1.1 Rapid 3­day system for Caco­2 monolayers: FITC­insulin permeation
In order to improve the aforementioned issues, a study by Gupta et. al investigated the permeabil­
ity of insulin using a modified rapid Caco­2 cell system [6]. In the article, the Caco­2 monolayers
were grown for three days in a serum­free growth medium, with added supplements such as butyric
acid, hormones, growth factors, and other metabolites. These served to regulate the differentia­
tion of intestinal epithelial cells. The transport across the modified 3­day Caco­2 monolayers of
three different polypeptides, including FITC­labeled insulin, was assessed at different loading con­
centrations. The apparent permeability coefficient, Papp , was then calculated using the equation:
Papp = dQ dQ
dt × A×C0 . dt which expresses the number of solutes transported across the Caco­2
1

barrier at the time dt, whilst C0 is the solute concentration in the apical compartment at time zero.
A represents the cross­sectional area of the epithelium in contact with the apical solution [6].
The article determined that the total amount of insulin transported across the Caco­2 monolayer
was directly proportional to the loading dose in the apical chamber, meaning transport exhibited
dose­dependency. This was determined by measuring the fluorescence of samples collected from
the basolateral chamber at different time points. In regards to the 3­day rapid system itself, it was
determined that the model had comparable P­glycoprotein activities and expressed similar levels
of metabolic enzymes (such as brush border peptidases) as compared to the 21­day conventional
model and in vivo conditions. In addition, the system seemed to provide more physiologically
relevant tight junctions with lower TEER values. In view of this, one can rationally propose the
usage of the novel 3­day rapid Caco­2 monolayer system as an alternative protocol for investigating
the permeability of oral insulin. In addition to physiological accuracy, it has various advantages,
such as the potential for saving time and resources, along with the possibility of performing high
throughput screening.

3.2 In vitro dissolution models

Dissolution studies are another highly relevant type of in vitro study for the investigation of a novel
oral insulin formulation. These studies evaluate how much of an orally administered dose will be
absorbed across the intestinal mucosa when it is released into a medium simulating GIT fluids and
physiological parameters. Performing these tests allows for a better understanding and prediction
of the possible behaviour of a new oral drug in in vivo studies [34, 35, 36].
From a technical point of view, the dissolution process begins when an oral drug formulation
changes its composition from its original form to a more molecularly distributed one. When it
comes to the dissolution of solid oral drug formulations, the process is divided into; the breakdown
of formulation into small particles, de­aggregation of the formulation, and, reaching the systemic
circulation [37]. The dissolution rate is affected by many factors, such as the surface area, the
diffusion layer thickness, the amount of drug in solution at a given time, and the saturated solu­
bility in the bulk medium [36]. In order to maximise the physiological accuracy in a dissolution
study, it is important to also consider the type of apparatus, agitation speed, temperature, media
composition, media volume, and choosing between fasted or fed state. Nonetheless, the two most
crucial parameters are the media volume and composition [35, 38].
Media volume is normally established by the fluid volume that simulates sink conditions (often­
times a range between 500­1000 mL) and the equilibrium solubility of the active pharmaceutical
ingredient (API) [38]. Development of biorelevant dissolution media has been made to simulate
GIT characteristics of common animal models used in preclinical studies of oral drug development.
Additionally, food intake also affects the dissolution of an oral drug. Hence, it must be considered
if the drug should be tested on fed­ or fasted­state media conditions [37].

Group 1: Oral insulin 7

3.2.1 Dissolution testing using a two­step dissolution model
Previous dissolution models typically relied on one­step dissolution models (thus only simulating
individual media fluids), which does not accurately resemble the real gastrointestinal (GI) transit
effect. As a result, implementing a two­step dissolution model is more precise, since the drug
formulation is exposed to changing GI fluids i.e. the stomach and intestinal fluid, respectively
[35]. Thusly, exposing the drug to the acidic environment of the stomach is necessary to measure
its integrity before using more neutral fluids to simulate the intestinal conditions ­ especially since
the release would primarily take place in the small intestine [39].
The change of the dissolution medium at a certain point during the studies is what defines a two­
step dissolution model. This can be done in two ways: by medium addition or by complete medium
replacement. Medium addition consists of exposing the dosage form to an acidic environment.
Afterward, a concentrated buffer solution is added to adjust the contents of the dissolution vessel
to a neutral pH, where the formulation will break down for the dissolution of the drug to proceed.
Medium replacement works by exposing the drug to an acidic medium followed by the removal
of the dosage unit from the vessel and subsequently submerging in a new dissolution vessel with
a neutral buffer medium [39].
Thus, this report proposes using a two­step dissolution model that considers carefully chosen
biorelevant media to simulate the gastric and intestinal fluids prior to conducting in vivo stud­
ies. Moreover, the dissolution should be tested in both fed­ and fasted­state biorelevant media to
investigate the effect of food intake on the formulation release. In addition, it is important to note
that the development of fed­state biorelevant media for canines and rats also depends on the type
of ingested meal. An article published by Grignard et al. describes the composition and physic­
ochemical properties of simulated media of fasted gastric and intestinal fluids in two common
animal models (rats and dogs, respectively) used for preclinical studies (cf. tables 3.1 and 3.2)
[40].

cFaSSGF cFaSSIF
pH 1.5 6.8
Sodium taurodeoxycholate (0.1 mM), Sodium taurocholate (5.00 mM),
Bile salts
sodium taurocholate (0.1 mM) sodiumtaurodeoxycholate (5.00 mM)
Phosphatidylcholine (0.025 mM), Phosphatidylcholine (1.25 mM),
Phospholipids
lysophosphatidylcholine (0.025 mM) lysophosphatidylcholine (1.25 mM)
Fatty acids Sodium oleate (0.025 mM) Sodium oleate (1.25 mM)
Hydrochloric acid, Sodium dihydrogen phosphate monohydrate (28.65 mM),
Buffer, cations, salts
sodium chloride(14.5 mM) sodium hydroxide (28 mM),sodium chloride (59.63 mM)
Enzymes Pepsin (600 U/h), lipase (190 U/h)
Table 3.1: Physiological parameters of canine fasted­state simulated gastric fluid (cFaSSGF) and canine fasted­state
simulated intestinal fluid (cFaSSIF). Data source: [40].

rFaSSGF rFaSSIF
pH 3.9 6.0
Bile salts 4 mM taurocholic acid 50 mM taurocholic acid
Phospholipids 0.2 mM phosphatidylcholine 2.2 mM phosphatidylcholine
0.02 M acetic acid;
0.1 M acetic acid;
Buffer, cations, salts 0.02 M sodium dihydrogen
0.1 M sodium dihydrogen phosphate
phosphate, 0.02 M sodium hydroxide
Enzymes Pepsin (1.2 μg/h), lipase (activity 44.3 U/h)
Table 3.2: Physiological parameters of rat fasted­state simulated gastric fluid (rFaSSGF) and rat fasted­state simulated
intestinal fluid (rFaSSIF). Data source: [40].

8 Group 1: Oral insulin

3.2.2 Dissolution testing using a TIM­1 model
The TNO gastrointestinal model, also known as the TIM­model, is a complex dissolution transfer
model that simulates the dynamic conditions of the upper GIT of humans, applying several physio­
logical parameters in combination with specific settings, such as the species, age, and meal­related
conditions [41].
One of the most prevalent models is TIM­1 which consists of four compartments. These represent
the stomach, duodenum, jejunum, and ileum, respectively. Hence, the model is able to simulate
several physiological conditions, for instance gastric emptying, intestinal transit times, changes in
pH values, and dynamics of mixing [42, 41, 43]. A graphic representation of this model can be ob­
served in Fig. 3.2. The TIM­1 model have shown high in vitro­in vivo correlation (IVIVC), when
comparing the results of jejunal absorption profiles with in vivo results. This makes the TIM­1
model an excellent dissolution model for testing the release of oral drugs in the upper GIT [42,
41, 43]. Other advantageous characteristics of the TIM­1 model includes filtration units which
are connected to the intestinal compartments, the ability to sample the GI lumen for dissolution
analysis, peristaltic valves pumps for controlling chyme transfer, and control of pH values by hy­
drochloric acid of sodium bicarbonate addition. The importance of the latter is elaborated upon in
Sec. 3.3.

Figure 3.2: TIM­1 model scheme that displays the dynamic nature by virtue of the technical setup of this dissolution
model. A: Stomach compartment; B Pyloric sphincter; C: Duodenum compartment; D: Perilstatic valve; E: Jejunum
compartment; F: Peristaltic valve; G: Ileum compartment; H: Ileo­caecal sphincter; I: Stomach secretion; J: Duodenum
secretion; K: Jejunum/ileum secretion; L: Pre­filter; M: Semi­permeable membrane; N: Filtrate pump; P: pH electrodes;
Q: Level sensors; R: Temperature sensor; S: Pressure sensor. Source of diagram: [42].

3.3 In vitro pH studies

pH is one of the most impactful factors that influences cellular functions. This parameter regulates
the activity of various enzymes and alters the binding affinity of hormones and neurotransmitters
[44], meaning the pH has a vital role in regulating cell functions. In metabolic pathophysiological
conditions, such as diabetes, the pH of interstitial fluids decreases, resulting in a more acidic GI
environment. This must be considered for the administration of oral insulin, which is most soluble
at acidic pH­values.
In addition, the insulin conformation is highly influenced by its concentration and surrounding
pH. Insulin has three structural forms: monomeric, dimeric, and hexameric. In humans, the drug
is synthesized as a monomer inside β­cells but once the concentration increases, it tends to form
dimers. When there is a sufficient presence of zinc (10mM) and favorable pH­conditions (pH∼6.0),

Group 1: Oral insulin 9

the monomers assemble into hexamers, which are stored in the body. Once the β­cell secretes the
insulin in its hexameric form, it diffuses into the bloodstream due to a lower concentration gradient.
Subsequently, the insulin dissociates into monomers. As this is the active form of the protein, it
can bind to its appropriate receptors to trigger specific physiological responses [45].
Besides the changes in structural conformations, insulin is shown to form aggregates under certain
conditions depending on the pH. These include amyloid­like fibrils and spherulites, which form
under acidic conditions, and particulates, which arise near the isoelectric point of insulin (pH∼5.3)
[46]. This might be problematic, as health authorities are increasingly concerned about these pro­
tein aggregates due to their association with an increased risk of immunogenic events. Besides
this, it can lead to a decrease in absorption, hence leading to poor glycemic control, causing higher
dosage requirements [47]. Due to the phenomena described, it is crucial to investigate the isoelec­
tric point (pI) of the novel drug and how pH will affect its folding and structure.
To determine the pI of the insulin, an Isoelectric Focusing Electrophoresis (IEF) should be con­
ducted. This is a precise and simple technique which allows the separation of amphoteric com­
pounds with high resolution in a medium containing a stable pH gradient. Using electrophoresis,
you generate a pH gradient by adding 50­100 carrier ampholytes with slight differences in pKa
values. When adding the protein, i.e. insulin, it will migrate until a point where the pH of the gel
reaches the protein’s pI, leading the protein to become neutral [48].
In regards to the structural dependence of the pH and formation of aggregates, this can be in­
vestigated using circular dichroism (CD) spectroscopy. CD is a powerful method that rapidly
determines the secondary structure and folding tendencies of proteins, which allows you to do
high­throughput testing of several samples using only low concentrations of the protein in physi­
ological buffers [49]. Hence, you can investigate the changes in the secondary structure of insulin
by adjusting the pH of the various samples. By choosing values ranging from pH 5 to 8, one can
beneficially investigate how the pH in the small intestine and bloodstream will affect the drug,
respectively [50].

3.4 In vitro toxicity testing

During preclinical studies, it is essential to not only consider the absorption and efficacy of the
drug but its safety as well. Hence, toxicological studies should be performed during the in vitro
tests too. The reasoning for this also lies in the ethics, as the three R’s should be considered before
conducting the subsequent in vivo studies, i.e. whether animal testing can be reduced, refined, and
replaced. For this, in vitro cytotoxicity tests are recommended although insulin as a subcutaneously
administered compound has been proven to be safe. These results are based on subcutaneous
administration, and the safety profile of the drug might change drastically depending on the change
in formulation (especially considering any added excipients etc.) [51].

3.4.1 Cytotoxicity testing in Caco­2 cells

As previously stated, Caco­2 cells monolayers are highly versatile due to the similarity to actual
human endothelial cells. Besides permeability testing, they can be applied to toxicology tests ­
particularly cytotoxicity assays to investigate if the compound induces cell death. When perform­
ing such studies, cell counting kits are often utilized. In this context, a CCK­8 assay can be used.
CCK­8 assays are robust and sensitive colorimetric assay that examines the viability of cells and
cytotoxicity of the drug. For the assay, Caco­2 cells are incubated for 24 hours. Subsequently,
the medium is changed to fresh medium containing different concentrations of the drug, i.e. the
oral insulin formulation. Following an additional day of co­cultivation, another medium change
containing 10% CCK­8 reagent is added to the cell monolayer [52]. Afterward, the absorbance of
the cells can be determined at a wavelength of 450 nm to determine the cell viability [53]. Thus,
you obtain a clear indication of the degree of cytotoxic impact of the oral insulin.

10 Group 1: Oral insulin

4 In vivo assays
In vivo tests are an essential part of the preclinical testing phase, as it is the only possible way
to evaluate a compound’s effect on the complete organism, examining both absorption, distribu­
tion, metabolism, and excretion (ADME) effects [54]. Animal studies are needed to find specific
pharmacokinetic (PK) properties of the formulation in question, which is fundamental when estab­
lishing the final dosage to be administered in first­in­human trials. Furthermore, it is a requirement
from the authorities to conduct in vivo toxicity studies when changing the route of administration
of a drug [55].
An important factor to consider when performing in vivo studies is the species selection. The most
frequently­employed species for in vivo testing include mice, rats, dogs, and nonhuman primates,
which all have different physiological characteristics [56]. According to the ICH S6 guidelines,
selecting relevant species when evaluating the efficacy of a drug is crucial. For a biopharmaceuti­
cal a relevant species is defined as an animal which expresses the relevant receptor needed for the
test compound pharmacological to be active. When evaluating the toxicity of a drug, the guide­
lines also states that only performing safety tests in one relevant species can be justified if the
biopharmaceutical examined has a well­known biological function, such as insulin [57].

4.1 In vivo pharmacology studies

In order to find the correct dosage, in vivo PK experiments are performed to determine properties
of the drug. Since the PK properties of insulin are well­established for subcutaneous injections,
the main objective of PK studies of the novel oral formulation of insulin would be to calculate
the relative bioavailability. Oftentimes, it can be challenging to measure the bioavailability of a
biopharmaceutical, since these can be difficult to detect in the blood. For insulin, however, the
bioavailability can be measured indirectly by measuring the blood glucose levels, as these are a
function of the insulin­levels.
A study by Liu et al. investigated the bioavailability of an orally administered insulin­filled nanopar­
ticle by measuring the blood glucose level of streptozotocin­induced diabetic rats after certain time
points and calculated the relative bioavailability by comparison with subcutaneous injection [58].
They calculated a relative bioavailability of 6%. If a likewise bioavailability was found for the
novel oral insulin formulation in this report, this would mean that it is necessary to administer a
16.6 times higher insulin dose for oral administration compared to subcutaneous. Despite a signif­
icantly lower bioavailability, it will not be an issue if insulin is produced inexpensively.
Overall, one must be critical when using glucose as the sole biomarker for insulin uptake, as this is
not solely dependent on insulin. Blood glucose levels can be further influenced by physiological
factors, such as stress or unrelated health conditions. Hence, it would be advantageous to measure
the actual insulin that reaches systemic circulation to support the findings, which can be done by
retrieving the plasma from rats and measuring the amount of insulin over time. This was done in
a study by Jørgensen et al., who measured blood glucose levels as well as the insulin levels in the
plasma when investigating the possibility of administering insulin orally by utilising microcontain­
ers. Additionally, it could be beneficial to measure additional biomarkers, such as the hemoglobin
A1c (HbA1c) levels, since this is used as a biomarker for hyperglycemia [59].

Group 1: Oral insulin 11

4.2 In vivo toxicity studies
In order to comply with the regulatory needs to progress onto clinical trials, toxicological in vivo
tests must be conducted. To reiterate, since insulin is a biopharmaceutical for which there is exten­
sive data and experience in clinical settings. Hence, the testing requirements are reduced, as testing
in one species in stead of two is deemed satisfactory, if adequate information can be obtained from
literature on e.g. the no­observed­adverse­effect level (NOAEL) of insulin [57].
When conducting in vivo tests, one must follow the guidelines issued by the Organisation for Eco­
nomic Co­operation and Development (OECD). For this, two tests are required: one investigating
acute toxicity and one that investigates chronic toxicity, respectively.
Acute toxicity is generally tested by administering a one­time dose of the drug in question to mul­
tiple rats. 14 days after the administration, different endpoints are studied to assess dose­related
differences between the rats. For this, differences in the relative weight of different body parts,
such as the brain, heart, liver, thymus, and spleen are commonly examined. To ensure the safety
of the novel oral insulin, the acute toxicity testing should show no difference in the relative weight
of the investigated endpoint when compared to the control rats [60]. Additionally, histological in­
vestigations of morphological changes in specific organs can be conducted. To assess the chronic
toxicity, a 28­day sub­chronic toxicity study should be performed. This can be achieved by ad­
ministering a repeated dose of the compound in question over a period of 28 days. Subsequently,
endpoints are evaluated in the same manner as described for acute toxicity testing [61].
Another aspect of the toxicity studies should focus on examining the immunotoxicity to ensure that
the new oral formulation does not generate an acute immunological response. This can be done
by analysing the serum samples collected from rats and detecting the presence of drug­related
antibodies by using a quantitative enzyme­linked immunosorbent assay (ELISA). For instance,
this was utilised in a chronic toxicology study of basal insulin by Byrd et al [62].

12 Group 1: Oral insulin

5 Discussion of selected methods
The holistic approach aims to give an overview of the useful preclinical tests that can provide
thorough insight into the novel insulin formulation. This is imperative, considering the many
unsuccessful attempts to make an oral insulin tablet.
The emphasis of this report has been the in vitro testing of dissolution and cell permeability. Al­
beit stated that the novel insulin formulation is not degraded in the stomach, this study still recom­
mends exposing it to the stomach’s acidic environment, as it was evaluated to offer a more precise
simulation of the changing pH in the GIT and ensure better IVIVC. This study recommends the
combination of both a two­step dissolution study along with a TIM­1 study. This choice was made
since the focus of the two­step dissolution study is to evaluate the effects on the drug release de­
pendent on the changing environment of the gastric and intestinal fluids. However, it does not
consider the effects of the real movement of the compound through the different regions of the
GIT. This is where the TIM­1 model can supplement the two­step model, due to its inherent ability
to simulate the movement. Additionally, the TIM­1 transfer model can simulate other physiolog­
ical parameters related to e.g. species, age, and meal­related variables. Despite this, it still has
certain shortcomings, such as the lack of intestinal mucosa and a relatively poor representation of
the actual fluid distribution in the small pockets of the small intestine [41, 42, 43]. A possible way
to improve upon this is to use intestinal cell lines, such as porcine intestinal tissue segments [43].
These two methods combined are expected to give sufficient insight into the dissolution of the
compound. However, further knowledge is gained by combining the TIM­1 with the TIM­2 model,
which focuses on simulating the large intestine [42]. Even though this would generate more data,
it is not deemed necessarily relevant for this study, since it is assumed the dissolution taking place
in the large intestine is minimal. Therefore, an extension of the study with the TIM­2 model would
presumably lead to extra costs and efforts, without yielding valuable insight. However, this report
argues that the dissolution testing should be further accompanied by independent pH tests, consid­
ering the immense impact that pH has on the conformation and biological function of insulin, cf.
Sec. 3.3.
The dissolution studies are compulsory, as only dissolved drug can be absorbed to have a thera­
peutic effect. Seeing as the small intestine is the primary location of absorption by virtue of its
large surface area, thorough testing of the endothelial cell permeability was also considered es­
sential. For this, the use of Caco­2 cells was proposed due to the presence of efflux transporters,
and relevant enzymatic activities, that resemble the actual intestinal environment to a large extent.
Specifically, the rapid 3­day protocol was suggested due to its improved tightness compared to
conventional 21­day cell lines. Moreover, since it is less time­ and resource­consuming, it gives
the option of performing high throughput testing. However, it is crucial to be critical of results
obtained in the 3­day protocol, since a shorter cultivation times would very likely result in lower
TEER­values.
Nonetheless, regardless of the protocol, Caco­2 cells are unable to produce a mucus layer, which
naturally acts as a gel filtration membrane that prevents the absorption of large molecules. To com­
pensate for this, one can favorably incorporate the use of microfluidics­based techniques, such as
intestine­on­a­chip (IOAC) technology [63]. When using Caco­2 cells on either 2D or 3D IOAC
models, one can simulate the continuous flow of fluids, peristaltic motions, the occurrence of
microvilli­structures, and the production of a mucosal layer, unlike the conventional, static Caco­2
cell models [64, 63]. However, even though there is huge potential in utilising IOAC technology,
there is still a lack of sufficient evidence before their usage can be integrated during preclinical

Group 1: Oral insulin 13

studies [65]. Therefore, this report refrains from recommending the use of IOACs, as the tech­
nique has yet to be well­acknowledged in the pharmaceutical industry.
In regards to in vivo testing, this study recommends only performing the necessary number of in
vivo investigations in order to comply with regulatory requirements. As mentioned in Sec. 3.4,
modern preclinical studies strive to practice the three R’s when performing animal experiments.
Besides the ethical concerns related, this also reduces costs to a certain degree. Therefore, this
study strongly suggests implementing PKPB models as in silico models to give an initial estima­
tion of the oral bioavailability. However, it is crucial to note that, despite the paradigm shift in
the pharmaceutical industry with the increased utilization of computational models, it is highly
unlikely that in silico models will be able to completely replace the need for animal studies. More­
over, the accuracy and predictive power of the models are limited by the lack of understanding of
the biological systems [66]. Additionally, the models depend on experimental data, as they must
be continuously improved as part of an iterative process. Hence, the final model is only as reliable
as the quality of the data input. This study suggests employing ACAT over CAT, since ACAT
estimates both the dissolution, permeation, and clearance, which are highly relevant pharmacoki­
netic parameters for this investigation. Moreover, even though they both yield information on the
dissolution­rate limited absorption, ACAT further evaluates the effect of formulation release rates
on oral pharmacokinetics [22]. Since ACAT is able to provide the most crucial pharmacokinetic
information, it is the only recommended in silico study in this report.
Before moving on to clinical trials, it is required to perform in vivo investigations of the bioavail­
ability, since animal models are still the only methods to obtain correct ADME information. To
calculate the relative bioavailability, it was suggested to orally administer insulin to rats and mon­
itor both blood glucose levels along with insulin levels in plasma. However, blood glucose levels
alone cannot directly be translated to bioavailability, since it has been proven that blood glucose
levels are also affected by various other parameters, such as adrenaline levels [67]. Therefore, it
is deemed necessary to also measure the serum insulin levels to most accurately calculate the oral
bioavailability in rats. Accurate calculations are essential to correctly estimate the appropriate dose
for first­in­human trials.
For the in vivo bioavailability studies, rats were recommended as a relevant animal species. One
might argue that it would be beneficial to conduct corresponding studies on a supplementary
species. For instance, dogs are generally a very compliant species and can be trained to swal­
low tablets. Contrarily, rats can not swallow tablets ­ instead, drugs are often administered using
oral gavage, which gives a less representative recreation of human oral conditions compared to
when using dogs. However, this study does not recommend using dogs, since it is evaluated that
rat studies would generate sufficient results. Therefore, it would be unethical to include dogs.
Additionally, using larger animal species is more expensive, which would only yield unnecessary
costs in this context. However, if inadequate or unexpected results are obtained from the in vivo
study, it would be necessary to further conduct studies in a non­rodent before being able to move
on to clinical trials.
This study also recommends performing acute and chronic in vivo toxicology screenings. One
might argue that these two studies alone will not demonstrate satisfactory proof of safety of the
novel formulation. This recommendation was, however, given based on the ICH guidelines S6,
which specifically states that ”Biopharmaceuticals that are structurally and pharmacologically
comparable to a product for which there is wide experience in clinical practice may need less ex­
tensive toxicity screening” [57]. Hence, it was evaluated that these two studies should be adequate
to examine if the novel formulation behaves like currently approved insulin products. Nonethe­
less, if the results from the toxicology study do not comply with earlier findings, it would be
recommended to conduct further animal studies in non­rodent species that share a higher degree
of similarity with humans.

14 Group 1: Oral insulin

In conclusion, all the preclinical tests recommended in this report are expected to supplement each
other throughout each step of the preclinical phase. All tests have been carefully chosen since they
have been predicted to yield all the needed information on the novel oral insulin formulation in
question to proceed onto clinical trials. However, even though this combination of tests theoret­
ically provides sufficient information, one must emphasize that the preclinical phase is a highly
dynamic and flexible process. Therefore, it is very likely that changes will happen when conduct­
ing the studies since unexpected test results evoke the need to performing additional tests or even
completely substituting certain tests.

Group 1: Oral insulin 15

Bibliography
[1] International Diabetes Federation. “Diabetes is “a pandemic of unprecedented magnitude”
now affecting one in 10 adults worldwide”. In: Diabetes Research and Clinical Practice 181
(2021). ISSN: 18728227. DOI: 10.1016/j.diabres.2021.109133.
[2] Yazid N Al Hamarneh et al. “Patient Assessment in Clinical Pharmacy”. In: Patient Assess­
ment in Clinical Pharmacy (2019). DOI: 10.1007/978-3-030-11775-7.
[3] Soren Havelund Welling. “Characterization of absorption enhancers for orally administered
therapeutic peptides in tablet formulations Applying statistical learning Soren Havelund
Welling”. In: (2016).
[4] Annette Schürmann and Hans­Georg Joost. “Diabetes Mellitus Diabetes Mellitus”. In: En­
cyclopedia of Molecular Pharmacology 512.58 (2021), pp. 432–441. DOI: 10.1016/B978-
0-323-67254-2.00255-2.
[5] Joanne Heade et al. “Synthesis and in vivo evaluation of insulin­loaded whey beads as an
oral peptide delivery system”. In: Pharmaceutics 13.5 (2021), pp. 1–18. ISSN: 19994923.
DOI: 10.3390/pharmaceutics13050656.
[6] Vivek Gupta, Nishit Doshi, and Samir Mitragotri. “Permeation of Insulin, Calcitonin and
Exenatide across Caco­2 Monolayers: Measurement Using a Rapid, 3­Day System”. In:
PLoS ONE 8.2 (2013). ISSN: 19326203. DOI: 10.1371/journal.pone.0057136.
[7] Luis Emilio García­Pérez et al. “Adherence to therapies in patients with type 2 diabetes”.
In: Diabetes Therapy 4.2 (2013), pp. 175–194. ISSN: 18696961. DOI: 10.1007/s13300-
013-0034-y.
[8] G.A. Harrison. “Insulin in alcoholic solution by the mouth”. In: Br. Med. J 22;2.3286 (1923),
pp. 1204–1205. DOI: 10.1136/bmj.2.3286.1204.
[9] Ruba Ismail and Ildikó Csóka. Novel strategies in the oral delivery of antidiabetic peptide
drugs – Insulin, GLP 1 and its analogs. June 2017. DOI: 10.1016/j.ejpb.2017.03.015.
[10] Guanyu Chen et al. Oral delivery of protein and peptide drugs: From non­specific formula­
tion approaches to intestinal cell targeting strategies. 2022. DOI: 10.7150/thno.61747.
[11] Gerardo P. Carino and Edith Mathiowitz. “Oral insulin delivery”. In: Advanced Drug Deliv­
ery Reviews 35.2­3 (1999), pp. 249–257. ISSN: 0169409X. DOI: 10.1016/S0169-409X(98)
00075-1.
[12] Chun Y. Wong, Jorge Martinez, and Crispin R. Dass. “Oral delivery of insulin for treat­
ment of diabetes: status quo, challenges and opportunities”. In: Journal of Pharmacy and
Pharmacology 68 (2016), pp. 1093–1108. ISSN: 20427158. DOI: 10.1111/jphp.12607.
[13] Ahmed Gedawy et al. “Oral insulin delivery: existing barriers and current counter­strategies”.
In: Journal of Pharmacy and Pharmacology 70.2 (2018), pp. 197–213. ISSN: 20427158.
DOI: 10.1111/jphp.12852.
[14] Thi Thao Linh Nguyen, Van An Duong, and Han Joo Maeng. Pharmaceutical formulations
with p­glycoprotein inhibitory effect as promising approaches for enhancing oral drug ab­
sorption and bioavailability. July 2021. DOI: 10.3390/pharmaceutics13071103.
[15] S. F. Zhou. Structure, function and regulation of P­glycoprotein and its clinical relevance
in drugdisposition. Vol. 38. 7­8. 2008, pp. 802–832. ISBN: 0049825070. DOI: 10.1080/
00498250701867889.
[16] Sarah Shugarts and Leslie Z. Benet. “The role of transporters in the pharmacokinetics of
orally administered drugs”. In: Pharmaceutical Research 26.9 (2009), pp. 2039–2054. ISSN:
07248741. DOI: 10.1007/s11095-009-9924-0.

16 Group 1: Oral insulin

[17] Søren H. Welling et al. “In silico modelling of permeation enhancement potency in Caco­
2 monolayers based on molecular descriptors and random forest”. In: European Journal
of Pharmaceutics and Biopharmaceutics 94 (2015), pp. 152–159. ISSN: 18733441. DOI:
10.1016/j.ejpb.2015.05.012. URL: http://dx.doi.org/10.1016/j.ejpb.2015.05.012.
[18] Min Wei et al. “HobPre: accurate prediction of human oral bioavailability for small molecules”.
In: Journal of Cheminformatics 14.1 (2022), pp. 1–10. ISSN: 17582946. DOI: 10.1186/
s13321-021-00580-6. URL: https://doi.org/10.1186/s13321-021-00580-6.
[19] Miguel Ángel Cabrera­Pérez and Hai Pham­The. Computational modeling of human oral
bioavailability: what will be next? June 2018. DOI: 10.1080/17460441.2018.1463988.
[20] Odilia Osakwe. Preclinical In Vitro Studies: Development and Applicability. Elsevier Inc.,
2016, pp. 129–148. ISBN: 9780128022207. DOI: 10.1016/b978-0-12-802220-7.00006-5.
URL: http://dx.doi.org/10.1016/B978-0-12-802220-7/00006-5.
[21] Balaji Agoram, Walter S. Woltosz, and Michael B. Bolger. “Predicting the impact of physio­
logical and biochemical processes on oral drug bioavailability”. In: Advanced Drug Delivery
Reviews 50.SUPPL. 1 (2001). ISSN: 0169409X. DOI: 10.1016/S0169-409X(01)00179-X.
[22] Louis Lin and Harvey Wong. “Predicting oral drug absorption: Mini review on physiologically­
based pharmacokinetic models”. In: Pharmaceutics 9.4 (2017), pp. 1–14. ISSN: 19994923.
DOI: 10.3390/pharmaceutics9040041.
[23] Olavi Pelkonen, Alan R. Boobis, and Ursula Gundert­Remy. “In vitro prediction of gas­
trointestinal absorption and bioavailability: An experts’ meeting report”. In: European jour­
nal of clinical pharmacology 57.9 (2001), pp. 621–629. ISSN: 00316970. DOI: 10.1007/
s002280100369.
[24] Dong Seok Yim, Soo Hyeon Bae, and Suein Choi. “Predicting human pharmacokinetics
from preclinical data: Clearance”. In: Translational and Clinical Pharmacology 29.2 (2021),
pp. 78–87. ISSN: 22890882. DOI: 10.12793/tcp.2021.29.e12.
[25] Lisa Cheng and Harvey Wong. “Food effects on oral drug absorption: Application of physiologically­
based pharmacokinetic modeling as a predictive tool”. In: Pharmaceutics 12.7 (2020), pp. 1–
18. ISSN: 19994923. DOI: 10.3390/pharmaceutics12070672.
[26] Heidi J. Einolf et al. “Physiologically based pharmacokinetic model predictions of panobi­
nostat (LBH589) as a victim and perpetrator of drug­drug interactions”. In: Drug Metabolism
and Disposition 45.12 (2017), pp. 1304–1316. ISSN: 1521009X. DOI: 10.1124/dmd.117.
076851.
[27] Pawel R. Kiela and Fayez K. Hishan. “The physiology of intestinal absorption.” In: Journal
of the Mount Sinai Hospital, New York 24.3 (1957), pp. 181–205. ISSN: 00999695. DOI:
10.1016/j.bpg.2016.02.007.Physiology.
[28] Noriyasu Kamei et al. “Mechanistic study of the uptake/permeation of cell­penetrating pep­
tides across a caco­2 monolayer and their stimulatory effect on epithelial insulin transport”.
In: Journal of Pharmaceutical Sciences 102.11 (2013), pp. 3998–4008. ISSN: 15206017.
DOI: 10.1002/jps.23708.
[29] E Le Ferrec et al. “In vitro models of the intestinal barrier. The report and recommendations
of ECVAM Workshop 46. European Centre for the Validation of Alternative methods.” In:
Alternatives to laboratory animals : ATLA 29.6 (2014), pp. 649–68. ISSN: 0261­1929. URL:
http://www.ncbi.nlm.nih.gov/pubmed/11709041.
[30] Andrew Arena, Jeanne Phillips, and Mark Blanchard. “AppNote: Drug transport assays in a
96­well system: reproducibility and correlation to human absorption”. In: (2022), pp. 1–10.
[31] Sanket M. Shah et al. Preclinical formulations: Insight, strategies, and practical consider­
ations. Oct. 2014. DOI: 10.1208/s12249-014-0156-1.
[32] Y. Yang et al. “Oral drug absorption: Evaluation and prediction”. In: Developing Solid Oral
Dosage Forms: Pharmaceutical Theory and Practice: Second Edition (2017), pp. 331–354.
DOI: 10.1016/B978-0-12-802447-8.00012-1.

Group 1: Oral insulin 17

[33] Yu Wang et al. “Transport mechanisms of polymannuronic acid and polyguluronic acid
across caco­2 cell monolayers”. In: Pharmaceutics 12.2 (2020). ISSN: 19994923. DOI:
10.3390/pharmaceutics12020167.
[34] Thaís dos Santos Paulino Soares et al. “Dissolution test for oral suspension: an overview
about use and importance”. In: Brazilian Journal of Pharmaceutical Sciences 58 (2022).
DOI: 10.1590/s2175-97902022e19423.
[35] Brijesh Shah and Xiaowei Dong. “Design and Evaluation of Two­Step Biorelevant Dissolu­
tion Methods for Docetaxel Oral Formulations”. In: AAPS PharmSciTech 23.5 (July 2022),
p. 113. ISSN: 1530­9932. DOI: 10.1208/s12249-022-02256-2.
[36] Ragna Berthelsen, Anette Müllertz, and Thomas Rades. “Evaluating Oral Drug Delivery
Systems: Dissolution Models”. In: 2016, pp. 753–771. DOI: 10.1007/978-1-4939-4029-
5{\_}24.
[37] Disha Mehtani et al. “Dissolution Profile Consideration in Pharmaceutical Product Devel­
opment”. In: Dosage Form Design Considerations. Elsevier, 2018, pp. 287–336. DOI: 10.
1016/B978-0-12-814423-7.00009-5.
[38] Raafat Fahmy and Marilyn N. Martinez. “Primer on the Science of In Vitro Dissolution Test­
ing of Oral Dosage Forms and Factors Influencing its Biological Relevance”. In: Dissolution
Technologies 26.1 (2019), pp. 14–26. ISSN: 1521298X. DOI: 10.14227/DT260119P14.
[39] Henry Zhao et al. “Practical considerations for the development of a robust two­step dis­
solution test for enteric­coated immediate­ and extended­release solid oral dosage formu­
lations”. In: Dissolution Technologies 18.1 (2011), pp. 6–10. ISSN: 1521298X. DOI: 10.
14227/DT180111P6.
[40] Elise Grignard et al. “Considerations for the development of in vitro dissolution tests to re­
duce or replace preclinical oral absorption studies”. In: European Journal of Pharmaceutical
Sciences 99 (Mar. 2017), pp. 193–201. ISSN: 09280987. DOI: 10.1016/j.ejps.2016.12.004.
[41] Mans Minekus. “The TNO Gastro­Intestinal Model (TIM)”. In: The Impact of Food Bioac­
tives on Health. Cham: Springer International Publishing, 2015, pp. 37–46. DOI: 10.1007/
978-3-319-16104-4{\_}5.
[42] Edmund S. Kostewicz et al. “In vitro models for the prediction of in vivo performance of oral
dosage forms”. In: European Journal of Pharmaceutical Sciences 57 (June 2014), pp. 342–
366. ISSN: 09280987. DOI: 10.1016/j.ejps.2013.08.024.
[43] Robert Havenaar and Susann Bellmann. “Results from a Validated in vitro Gastrointestinal
Model (TIM) used as input Data for in silico Modeling Give Highly Predictive Information
for the Human Situation”. In: Medical Research Archives 10.9 (2022). ISSN: 23751916.
DOI: 10.18103/mra.v10i9.3136.
[44] Yoshinori Marunaka. “Roles of interstitial fluid pH in diabetes mellitus: Glycolysis and
mitochondrial function.” eng. In: World journal of diabetes 6.1 (Feb. 2015), pp. 125–135.
ISSN: 1948­9358 (Print). DOI: 10.4239/wjd.v6.i1.125.
[45] Zhuo Fu, Elizabeth R Gilbert, and Dongmin Liu. “Regulation of insulin synthesis and se­
cretion and pancreatic Beta­cell dysfunction in diabetes.” eng. In: Current diabetes reviews
9.1 (Jan. 2013), pp. 25–53. ISSN: 1875­6417 (Electronic).
[46] Camilla Thorlaksen et al. “Subtle pH variation around pH 4.0 affects aggregation kinetics
and aggregate characteristics of recombinant human insulin”. In: European Journal of Phar­
maceutics and Biopharmaceutics 179.May (2022), pp. 166–172. ISSN: 09396411. DOI:
10.1016/j.ejpb.2022.09.001.
[47] Yui Ohno et al. “Investigation of factors that cause insulin precipitation and/or amyloid
formation in insulin formulations”. In: Journal of Pharmaceutical Health Care and Sciences
5.1 (2019), p. 22. ISSN: 2055­0294. DOI: 10 . 1186 / s40780 - 019 - 0151 - 5. URL: https :
//doi.org/10.1186/s40780-019-0151-5.

18 Group 1: Oral insulin

[48] Findley N Cornell. “Isoelectric focusing, blotting and probing methods for detection and
identification of monoclonal proteins.” eng. In: The Clinical biochemist. Reviews 30.3 (Aug.
2009), pp. 123–130. ISSN: 1838­0212 (Electronic).
[49] Norma J Greenfield. “Using circular dichroism spectra to estimate protein secondary struc­
ture.” eng. In: Nature protocols 1.6 (2006), pp. 2876–2890. ISSN: 1750­2799 (Electronic).
DOI: 10.1038/nprot.2006.202.
[50] Ekaterina Smirnova et al. “pH­responsive modulation of insulin aggregation and structural
transformation of the aggregates.” eng. In: Biochimie 109 (Feb. 2015), pp. 49–59. ISSN:
1638­6183 (Electronic). DOI: 10.1016/j.biochi.2014.12.006.
[51] Sonal Srivastava et al. “Chapter 2 ­ Principles for In Vitro Toxicology”. In: ed. by Alok
Dhawan and Seok B T ­ In Vitro Toxicology Kwon. Academic Press, 2018, pp. 21–43.
ISBN: 978­0­12­804667­8. DOI: https://doi.org/10.1016/B978-0-12-804667-8.00002-
X. URL: https://www.sciencedirect.com/science/article/pii/B978012804667800002X.
[52] Xin Fan et al. “Anti­Colon Cancer Activity of Novel Peptides Isolated from In Vitro Di­
gestion of Quinoa Protein in Caco­2 Cells.” eng. In: Foods (Basel, Switzerland) 11.2 (Jan.
2022). ISSN: 2304­8158 (Print). DOI: 10.3390/foods11020194.
[53] Weizhi Wang et al. “Improved oral delivery of insulin by PLGA nanoparticles coated with
5β­cholanic acid conjugated glycol chitosan.” eng. In: Biomedical materials (Bristol, Eng­
land) 16.6 (Oct. 2021). ISSN: 1748­605X (Electronic). DOI: 10.1088/1748-605X/ac2a8c.
[54] Meisam Omidi et al. “Characterization of biomaterials”. In: Biomaterials for Oral and Den­
tal Tissue Engineering. Elsevier, Jan. 2017, pp. 97–115. ISBN: 9780081009611. DOI: 10.
1016/B978-0-08-100961-1.00007-4.
[55] U.S. Department of Health Services and Human. “Guidance for industry and review staff
­ non clinical safety evaluation of reformulated drug products and products intended for
administration by an alternate route”. In: March (2008), pp. 1–8.
[56] Jeanine L. Bussiere. Species selection considerations for preclinical toxicology studies for
biotherapeutics. July 2008. DOI: 10.1517/17425255.4.7.871.
[57] European Medicine Agency. “EMA/CHMP/ICH/731268/1998 ICH guideline S6 (R1) on
preclinical safety evaluation of biotechnology­derived pharmaceuticals”. In: Committee for
medicinal products for human use (CHMP) ICH 6.June (2011), pp. 1–22. URL: https://
www.ema.europa.eu/en/documents/scientific- guideline/ich- s6r1- preclinical- safety-
evaluation-biotechnology-derived-pharmaceuticals-step-5_en.pdf.
[58] Liyao Liu et al. “Self­assembled lecithin/chitosan nanoparticles for oral insulin delivery:
Preparation and functional evaluation”. In: International Journal of Nanomedicine 11 (2016),
pp. 761–769. ISSN: 11782013. DOI: 10.2147/IJN.S96146.
[59] Timothy J. Lyons and Arpita Basu. “Biomarkers in diabetes: Hemoglobin A1c, vascular and
tissue markers”. In: Translational Research 159.4 (2012), pp. 303–312. ISSN: 18781810.
DOI: 10.1016/j.trsl.2012.01.009. URL: http://dx.doi.org/10.1016/j.trsl.2012.01.009.
[60] OECD/OCDE 407 OECD GUIDELINES FOR THE TESTING OF CHEMICALS Repeated
Dose 28­Day Oral Toxicity Study in Rodents INTRODUCTION. Tech. rep.
[61] OECD/OCDE 423 OECD GUIDELINE FOR TESTING OF CHEMICALS Acute Oral Toxicity­
Acute Toxic Class Method INTRODUCTION. Tech. rep. 2001.
[62] Richard A Byrd et al. “Chronic Toxicology Studies of Basal Insulin Peglispro in Rats and
Dogs: A Novel, PEGylated Insulin Lispro Analog with a Prolonged Duration of Action”. In:
Toxicologic Pathology 45.3 (Mar. 2017), pp. 402–415. ISSN: 0192­6233. DOI: 10.1177/
0192623317696283. URL: https://doi.org/10.1177/0192623317696283.
[63] Yunqing Xiang et al. “Gut­on­chip: Recreating human intestine in vitro”. In: Journal of
Tissue Engineering 11 (2020). ISSN: 20417314. DOI: 10.1177/2041731420965318.

Group 1: Oral insulin 19

[64] Hsih Yin Tan et al. “A multi­chamber microfluidic intestinal barrier model using Caco­2
cells for drug transport studies”. In: PLoS ONE 13.5 (2018), pp. 1–23. ISSN: 19326203.
DOI: 10.1371/journal.pone.0197101.
[65] Chao Ma et al. “Organ­on­a­Chip: A New Paradigm for Drug Development”. In: Trends in
Pharmacological Sciences 42.2 (2021), pp. 119–133. ISSN: 18733735. DOI: 10.1016/j.
tips.2020.11.009. URL: https://doi.org/10.1016/j.tips.2020.11.009.
[66] Jack W. Scannell et al. “Diagnosing the decline in pharmaceutical R&D efficiency”. In:
Nature Reviews Drug Discovery 11.3 (2012), pp. 191–200. ISSN: 14741776. DOI: 10.1038/
nrd3681.
[67] Hilda Wong et al. “The Effects of Mental Stress on Non­insulin­dependent Diabetes: De­
termining the Relationship Between Catecholamine and Adrenergic Signals from Stress,
Anxiety, and Depression on the Physiological Changes in the Pancreatic Hormone Secre­
tion”. In: Cureus (2019), pp. 1–8. DOI: 10.7759/cureus.5474.

20 Group 1: Oral insulin

Group 1: Oral insulin 21
Technical
University of
Denmark

Ørsteds Plads, Building 345C

2800 Kgs. Lyngby
Tlf. 4525 1700

https://www.healthtech.dtu.dk/english