Tumour regression and improved gastrointestinal tolerability from controlled release of SN-38 from novel polyoxazoline-modified dendrimers

Article abstract of DOI:10.1016/j.jconrel.2016.12.034

Irinotecan is used clinically for the treatment of colorectal cancer; however, its utility is limited by its narrow therapeutic index. We describe the use of a generation 5 L-lysine dendrimer that has been part-modified with a polyoxazoline as a drug delivery vehicle for improving the therapeutic index of SN-38, the active metabolite of irinotecan. By conjugating SN-38 to the dendrimer via different linker technologies we sought to vary the release rate of the drug to generate diverse pharmacokinetic profiles. Three conjugates with plasma release half-lives of 2.5?h, 21?h, and 72?h were tested for efficacy and toxicity using a mouse SW620 xenograft model. In this model, the linker with a plasma release half-life of 21?h achieved sustained SN-38 exposure in blood, above the target concentration. Control over the release rate of the drug from the linker, combined with prolonged circulation of the dendrimer, enabled administration of an efficacious dose of SN-38, achieving significant regression of the SW620 tumours. The conjugates with 2.5 and 72?h release half-lives did not achieve an anti-tumour effect. Intraperitoneal dosing of the clinically used prodrug irinotecan produces high initial and local concentrations of SN-38, which are associated with gastrointestinal toxicity. Administration of the 21?h release dendrimer conjugate did not produce a high initial C_max of SN-38. Consequently, a marked reduction in gastrointestinal toxicity was observed relative to irinotecan treatment. Additional studies investigating the dose concentrations and dose scheduling showed that a weekly dosing schedule of 4?mg SN-38/kg was the most efficacious regimen. After 4 doses at weekly intervals, the survival period of the mice extended beyond 70 days following the final dose. These extensive studies have allowed us to identify a linker, dose and dosing regimen for SN-38 conjugated to polyoxazoline-modified dendrimer that maximised efficacy and minimised adverse side effects.

Full text of DOI:10.1016/j.jconrel.2016.12.034

Tumour regression and improved gastrointestinal tolerability from controlled
release of SN-38 from novel polyoxazoline-modified dendrimers
Richard M. England, Jennifer I. Hare, Jennifer Barnes, Joanne Wilson,
Aaron Smith, Nicole Strittmatter, Paul Kemmitt, Michael J. Waring, Simon T.
Barry, Cameron Alexander, Marianne Ashford
PII:
DOI:
Reference:
S0168-3659(16)30869-0
doi:10.1016/j.jconrel.2016.12.034
COREL 8590
To appear in:
Journal of Controlled Release
Received date:
Revised date:
Accepted date:
28 September 2016
20 December 2016
29 December 2016
Please cite this article as: Richard M. England, Jennifer I. Hare, Jennifer Barnes,
Joanne Wilson, Aaron Smith, Nicole Strittmatter, Paul Kemmitt, Michael J. War-
ing, Simon T. Barry, Cameron Alexander, Marianne Ashford, Tumour regres-
sion and improved gastrointestinal tolerability from controlled release of SN-38
from novel polyoxazoline-modified dendrimers, Journal of Controlled Release (2016),
doi:10.1016/j.jconrel.2016.12.034
This is a PDF file of an unedited manuscript that has been accepted for publication.
As a service to our customers we are providing this early version of the manuscript.
The manuscript will undergo copyediting, typesetting, and review of the resulting proof
before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that
apply to the journal pertain.

ACCEPTED MANUSCRIPT
Tumour regression and improved gastrointestinal tolerability from
controlled release of SN-38 from novel polyoxazoline-modified dendrimers
Richard M. England,a,1,* Jennifer I. Hare,b,1,* Jennifer Barnes,d Joanne Wilson,c Aaron
Smith,c Nicole Strittmatter,d Paul Kemmitt,b Michael J. Waring,e Simon T. Barry,c Cameron
Alexander,f and Marianne Ashford.a
a. AstraZeneca, Pharmaceutical Sciences, Innovative Medicines, Silk Court Business Park, Macclesfield,
Cheshire, SK10 2NA, United Kingdom.
b. AstraZeneca, IMED Oncology, Alderley Park, Macclesfield, Cheshire, UK, SK10 4TG, United
Kingdom.
c. AstraZeneca, IMED Oncology, Li Ka Shing Centre, CRUK Cambridge Institute, Cambridge, CB2
0RE, United Kingdom.
d. AstraZeneca, Pathological Sciences, Drug Safety and Metabolism, Cambridge, CB4 0WG, United
Kingdom.
e. Northern Institute for Cancer Research, School of Chemistry, Bedson Building, Newcastle University,
Newcastle upon Tyne, NE1 7RU, United Kingdom.
f. School of Pharmacy, University of Nottingham, Nottingham, NG7 2RD, United Kingdom.
1Authors contributed equally to this work.
*Corresponding author E-mail addresses: Richard.England@astrazeneca.com, Jennifer.Hare@astrazeneca.com
1

ACCEPTED MANUSCRIPT
Graphical abstract
t1/2 = 2.5 h “Fast”
SN-38
t1/2 = 21 h “Medium”
t1/2 = 72 h “Slow”
Abstract
Irinotecan is used clinically for the treatment of colorectal cancer; however, its utility is limited by its
narrow therapeutic index. We describe the use of a generation 5 L-lysine dendrimer that has been
part-modified with a polyoxazoline as a drug delivery vehicle for improving the therapeutic index of
SN-38, the active metabolite of irinotecan. By conjugating SN-38 to the dendrimer via different linker
technologies we sought to vary the release rate of the drug to generate diverse pharmacokinetic
profiles. Three conjugates with plasma release half-lives of 2.5 h, 21 h, and 72 h were tested for
efficacy and toxicity using a mouse SW620 xenograft model. In this model, the linker with a plasma
release half-life of 21 h achieved sustained SN-38 exposure in blood, above the target concentration.
Control over the release rate of the drug from the linker, combined with prolonged circulation of the
dendrimer, enabled administration of an efficacious dose of SN-38, achieving significant regression
of the SW620 tumours. The conjugates with 2.5 and 72 h release half-lives did not achieve an anti-
tumour effect. Intraperitoneal dosing of the clinically used prodrug irinotecan produces high initial
and local concentrations of SN-38, which are associated with gastrointestinal toxicity. Administration
of the 21 h release dendrimer conjugate did not produce a high initial Cmax of SN-38. Consequently, a
marked reduction in gastrointestinal toxicity was observed relative to irinotecan treatment.
Additional studies investigating the dose concentrations and dose scheduling showed that a weekly
dosing schedule of 4 mg SN-38/kg was the most efficacious regimen. After 4 doses at weekly
intervals, the survival period of the mice extended beyond 70 days following the final dose. These
extensive studies have allowed us to identify a linker, dose and dosing regimen for SN-38 conjugated
to polyoxazoline-modified dendrimer that maximised efficacy and minimised adverse side effects.
Keywords
Nanomedicine; irinotecan; SN-38; dendrimer; colorectal cancer; drug delivery; therapeutic index.
2

ACCEPTED MANUSCRIPT
1. Introduction
One of the key considerations for cancer medicines remains establishing a therapeutic index [1],
since targeting key cellular processes can induce significant toxicity, and therapeutic margins are
typically narrow, or almost non-existent for some specific potent cytotoxic therapies.
Topoisomerases are ubiquitous enzymes that control DNA supercoiling and entanglements. They
cleave the DNA backbone, releasing DNA supercoils and re-ligating the cleaved DNA.
Topoisomerases are essential during the transcription and replication process, and topoisomerase
inhibitors that cause single or double strand cleavage (type I or II, respectively) are among the most
effective and most commonly used anticancer and antibacterial drugs [2]. Topotecan and irinotecan
are topoisomerase I inhibitors routinely used in the clinic in a range of different cancers. They are
water soluble semi-synthetic derivatives of the potent cytotoxic plant alkaloid camptothecin.
However, serious adverse effects, including myelosuppression and diarrhoea, have limited their
efficacy in the clinic.
In addition, irinotecan is a prodrug, which is hydrolysed to its more potent active metabolite, SN-38,
by carboxylesterases in the liver and tumour cells. This conversion is often inefficient in humans, and
leads to variability in SN-38 exposure [3]. Although SN-38 is ~100- to 1000-fold more potent than
irinotecan [3], its clinical use is limited by its poor water solubility and chemical instability of its
pharmacologically active lactone ring at pH >6. Alongside the narrow therapeutic index of irinotecan
and SN-38, these issues present further challenges for their use as anti-cancer therapeutics in the
clinic.
In principle, one way to improve therapeutic index is to increase drug exposure in diseased tissue,
without accumulation in other healthy tissues. Nanotechnology offers the possibility to achieve
3

ACCEPTED MANUSCRIPT
greater site specificity by changing the biodistribution of a drug. These drug delivery systems have
been shown to increase drug concentration at tumour sites, relative to off-target tissues. This is
often achieved by minimising the high initial drug concentrations in the plasma (Cmax) and allowing
tissue accumulation and retention via the enhanced permeability and retention (EPR) effect [4, 5].
The rapidly developing field of nanomedicines covers a range of nano-scale constructs, such as
liposomes, polymeric nanoparticles, polymeric micelles, various polymer conjugates and inorganic
particles.
Several nanomedicine delivery strategies have been investigated to improve the therapeutic index of
SN-38 and camptothecin analogues, by reducing accumulation in normal tissues while increasing
delivery to tumours. To date, SN-38 and camptothecin analogues have been delivered using
liposomes, polymeric nanoparticles, polymeric micelles and various polymer conjugates and
antibody drug conjugates [6-12].
Onivyde™ (Merrimack Pharmaceuticals), a liposomal irinotecan, has recently been approved for use
in metastatic pancreatic cancer and is currently being used in Phase II trials in gastric cancer.
Etirinotecan pegol (ONZEALD™, formerly NKTR-102; Nektar Therapeutics) is in advanced clinical
development for ovarian, breast, colon, and lung cancers [13, 14]. Both of these nano-formulations
provide greater area under the curve (AUC) for SN-38 than does irinotecan treatment in pre-clinical
tumours [14]. A camptothecin nanoparticle polymer conjugate, CRLX101 (Cerulean Pharma Inc.) is
currently in clinical trials in a number of cancers. CRLX101 has demonstrated an extended circulation
half-life and prolonged release of camptothecin in rat, dog, and human plasma [15], sustained
tumour pharmacokinetics and inhibition of target [16], and evidence of localisation in tumour rather
than healthy tissues in gastric cancer patients [17].
4

ACCEPTED MANUSCRIPT
All of these systems have demonstrated positive results in pre-clinical models and are progressing in
the clinic with promising activity. However, the ability to control the release of the payload from the
nano-carrier has not been explored, precluding the possibility of fine-tuning the release rate to
maximise the therapeutic index. The ability to explore the relationship between efficacy and toxicity
across a range of release rates generates insight that drives biology-focussed nanomedicine design
[18].
Dendrimer drug conjugates offer many advantages over other drug delivery systems, including
tuneable drug release. Additional benefits of the dendrimer platform include: control of size during
synthesis, near-monomodal molecular weight distributions, and reproducible synthetic sequences.
Dendrimers are significantly smaller (~7-15 nm) than polymeric nanoparticles prepared via other
routes. Nanomedicine systems of similar sizes have been shown to extravasate to a greater extent
and/or penetrate farther from the vasculature than do larger systems, which has been associated
with improved efficacy [19-21]. In addition, dendrimers have large numbers of surface groups
available for conjugation. These surface groups can be used for drug conjugation via different linker
chemistries, without significantly impacting solubility, or to modify the outer layer of the dendrimer.
This modification controls surface charge and provides a corona to prevent protein adsorption,
thereby extending the circulation time of the dendrimer conjugate. PEGylation has been widely
employed with polyamidoamine (PAMAM) [26-29] and poly(L-lysine) dendrimers [30-33] for surface
modification. There is now significant diversity in the dendrimer-based nanomedicines being
explored pre-clinically, including amphiphilic dendrimers forming nanomicelles [22], dendrimer-
assisted metal nanocomposite particles for CT/MR imaging [23], lactoferrin-containing dendriplexes
[24], and doxorubicin-conjugated dendrimers for pulmonary delivery [25].
We have built on this strategy using a biodegradable fifth generation L-lysine dendrimer, part-
modified with a polyoxazoline [34], to deliver the potent topoisomerase I inhibitor SN-38.
5

ACCEPTED MANUSCRIPT
Recognising the importance of controlled drug release in designing nanomedicines and optimising
therapeutic index [35-40], we designed a range of linker chemistries to tune the release rate of SN-
38, and explored the synthesis, characterisation and therapeutic application of three dendrimer-
SN38 (Dend-SN38) conjugates in vitro and in vivo.
2. Materials and methods
Materials
All reagents and solvents were purchased from Sigma-Aldrich and used as received with exception of
SN-38 (TCI Europe N.V.), N,N′ -Disuccinimidyl carbonate & 2-chloro-4,6-dimethoxy-1,3,5-triazine
(Chem-Impex International). Irinotecan was purchased from CarboSynth (UK).
Methods
HPLC-UV was performed on an Agilent 1100 fitted with an Atlantis dC18 5 μm 4.6 x 150 mm column
with an acetonitrile/water gradient + 1% TFA modifier. See Supporting Information Table S1 for
HPLC-UV run parameters. The data were analysed using Thermo Scientific™ Atlas Chromatography
Data System (CDS) software. GPC was performed on a Malvern TDA302 using a Tosoh Bioscience
TSKgel GMPWxl column and an eluent of a 40: 10 mM NaNO3: NaH2PO4 + 10% MeOH in water.
Samples were made to 2 mg/ml in the eluent. The system calibration was performed using a single
poly(ethylene oxide) standard (16 100 g mol−1) and absolute molecular weights determined using the
refractive index and light scattering signals. Dynamic light scattering was performed on a Malvern
Zetasizer® Nano ZS instrument with back scattering detector (173°, 633 nm laser wavelength). The
dispersant RI and viscosity were assumed to be that of water (n = 1.59 and η = 0.888 mPa·s). The
sample RI was 1.59 and the temperature was set at 25°C. The hydrodynamic diameter (DH) was
reported as the volume-weighted average after a minimum of twelve measurements per sample and
6

ACCEPTED MANUSCRIPT
was calculated by the software. Samples were made to concentrations of 5 mg/ml in PBS at pH 7.4
and were filtered using a 0.2 μm syringe filter prior to measurement. Data was obtained using
Malvern Zetasizer software version 6.21J.
4-(4,6-Dimethoxy-1,3,5-triazin-2-yl)-4-methylmorpholinium tetrafluoroborate (DMTMM.BF4) was
synthesised according to methods reported previously [41].
G5-PLL[PMOx]x[NH2]y. This product was synthesized according to methods previously described [34].
Mn = 75.7 kg/mol, PDI = 1.17 from triple detection size exclusion chromatography.
G5-PLL[PMOx]x[Azide]y. G5-PLL[PMOx][NH2] (750 mg, 9.78 μmol) was dissolved in anhydrous DMF (5
mL) at 45°C. The solution was cooled to room temperature. A pre-mixed solution of 2-azidoacetic
acid (93 μL, 1.25 mmol), N-(3-Dimethylaminopropyl)-N′ -ethylcarbodiimide hydrochloride (EDC) (240
mg, 1.25 mmol) and 4-(Dimethylamino)pyridine (DMAP) (15 mg, 0.12 mmol) in anhydrous DMF (5
mL) was added to the polymer solution and stirred overnight at room temperature. The reaction
mixture was added drop-wise to rapidly stirring diethyl ether (200 mL) to form a white precipitate
which was collected by filtration and dried under vacuum for 2 hours at room temperature. The solid
was then taken up in deionised water (50 mL), passed through a 0.45 μm nylon filter cup and
purified by VivaFlow® cassette in deionised water (30 kDa MWCO). The title compound was
obtained as a white solid after freeze-drying (675 mg, 87 %). 1H NMR (400 MHz, D2O) 1.12 – 1.86 (br,
367H), 1.89 – 2.3 (br, 1500H), 3.19 (br, 136H), 3.56 (br, 1940H), 3.92 (br, 111H), 4.04 – 4.44 (br,
130H).
6-amino-N-(prop-2-ynyl)hexanamide. Boc-6-aminohexanoic acid (1.0 g, 4.32 mmol) was dissolved in
anhydrous DCM (25 mL) in a reaction tube purged with nitrogen. EDC (1.34 g, 8.64 mmol) was added
as a solid and stirred until dissolved. DMAP (0.1 g, 8.65 mmol) was added followed by
7

ACCEPTED MANUSCRIPT
propargylamine (0.831 mL, 12.97 mmol) dropwise via Gilson pipette. The reaction was stirred
overnight at room temperature before diluting with further DCM (25 mL) and washing with 0.1M HCl
(2 x 25 mL), saturated NaHCO3 (2 x 25 mL) and saturated brine (25 mL). The organic phase was dried
over magnesium sulfate, filtered and evaporated under vacuum to obtain the title compound as a
white solid (0.632 g, 55 %).1H NMR (CDCl3, 400 MHz): δ=1.36 (m, 2H), δ=1.44 (s, 9H), δ=1.50 (m,
2H), δ=1.67 (m, 2H), δ=3.11 (q, 2H), δ=4.05 (dd, 2H), δ=4.46-4.61 (b, 1H), δ=5.56-5.69 (b, 1H).
6-amino-N-(prop-2-yn-1-yl)hexanamide.HCl. 6-amino-N-(prop-2-ynyl)hexanamide (0.60 g, 2.24
mmol) was dissolved in methanol (5 mL) in a round bottom flask fitted with a gas outlet. HCl (1.68
mL, 4M in dioxane) was added drop-wise and the reaction was stirred at room temperature for 6
hours. The solvents were evaporated under vacuum and the residue was washed with DCM (25 mL).
The remaining solid was dried under vacuum at room temperature overnight to give the title
compound as a pale brown solid (0.39 g, 85 %). 1H NMR (400MHz, d-DMSO) 1.22 – 1.37 (2H, m),
1.49 (2H, m), 1.53 – 1.62 (2H, m), 2.09 (2H, t), 2.64 – 2.82 (2H, m), 3.06 (1H, t), 3.83 (2H, dd).
tert-butyl methyl(3-oxo-3-(prop-2-yn-1-ylamino)propyl)carbamate. This compound was
synthesized as above using boc-sarcosine (1.0 g, 5.29 mmol). The title compound was obtained as a
white solid (0.59 g, 49 %). 1H NMR (CDCl3, 400 MHz): δ=1.48 (s, 9H), δ=2.23 (t, 1H), δ=1.48 (s, 3H),
δ=3.86 (s, 2H), δ=4.07 (dd, 2H).
tert-butyl methyl(2-oxo-2-(prop-2-yn-1-ylamino)ethyl)carbamate.HCl. This compound was
synthesised according to the above method. The title compound was obtained as a pale brown solid
(0.38 g, 72 %). 1H NMR (d6-acetone, 400 MHz): δ=2.87 (s, 3H), δ=3.66 (s, 1H), δ=3.98 (m, 4H).
SN-38 6-oxo-6-(prop-2-ynylamino)hexylcarbamate (1a). SN-38 (75 mg, 0.19 mmol) was dissolved in
a mixture of THF (6 mL) and acetonitrile (6 mL) with N,N-Diisopropylethylamine (DIPEA) (0.100 mL,
8

ACCEPTED MANUSCRIPT
0.57 mmol) in a 10 mL reaction vial purged with nitrogen. bis(2,5-dioxopyrrolidin-1-yl) carbonate
(196 mg, 0.76 mmol) was added directly as a solid. The reaction was monitored by HPLC and when
full conversion of the SN-38 was observed (~30 min) the reaction mixture was diluted with EtOAc (25
mL), and washed with 0.1M HCl (2 x 10 mL) and saturated brine (2 x 10 mL). The organic layer was
dried over MgSO4, filtered and evaporated to afford the intermediate. The intermediate was
dissolved in DCM (15 mL) before DMAP (9.34 mg, 0.08 mmol), and 1a (58.7 mg, 0.29 mmol)
dissolved in DMF (2 mL) were added followed by DIPEA (33 μL, 0.19 mmol). The reaction mixture
was stirred at room temperature under nitrogen until the reaction had gone to completion
(measured by LCMS, ~10 min). The solvent was evaporated and the crude product was purified by
flash silica chromatography with an elution gradient of 10-80% 3:1 EtOAc: EtOH in heptane. The title
compound was obtained as a pale yellow solid after evaporation. (85 mg, 76 %). 1H NMR (400 MHz,
CDCl3) 1.04 (4H, t), 1.31 – 1.49 (8H, m), 1.6 – 1.69 (3H, m), 1.72 (3H, dd), 1.79 – 2 (3H, m), 2.15 – 2.29
(4H, m), 2.51 – 2.63 (1H, m), 2.74 (2H, s), 2.88 (2H, d), 2.95 (2H, s), 3.16 (2H, q), 3.34 (2H, dt), 4.06
(3H, dd), 5.26 (3H, s), 5.28 – 5.37 (2H, m), 5.75 (1H, d), 7.60 (1H, dd), 7.63 (1H, s), 7.88 (1H, d), 8.21
(1H, d).
SN-38 methyl(2-oxo-2-(prop-2-ynylamino)ethyl)carbamate (1c). This compound was synthesized as
for 2a. The title compound was obtained as a pale yellow solid (76 mg, 73 %). 1H NMR (400 MHz,
CDCl3) 1.04 (3H, t), 1.42 (3H, dt), 1.54 (3H, s), 1.75 – 2 (2H, m), 2.26 (1H, s), 3.15 (3H, d), 3.30 (2H, s),
3.49 (1H, s), 3.76 (1H, d), 3.96 – 4.28 (4H, m), 5.24 (2H, s), 5.27 – 5.33 (1H, m), 5.75 (1H, dd), 7.61
(2H, s), 7.8 – 7.92 (1H, d), 8.22 (1H, d).
SN-38 hex-5-ynoate (1b). Hex-5-ynoic acid (23 μL, 0.21 mmol) was dissolved in anhydrous DMF (7
mL) in a nitrogen-purged reaction vial. EDC (73.3 mg, 0.38 mmol) was added as a solid, followed by
DMAP (46.7 mg, 0.38 mmol) and the mixture stirred for 5 minutes. 7-Ethyl-10-hydroxycamptothecin
(75 mg, 0.19 mmol) was added as a solid and the reaction stirred overnight. The reaction mixture
9

ACCEPTED MANUSCRIPT
was diluted with ethyl acetate (50 mL), and washed with 0.1M HCl (2 x 20 mL), and saturated brine
(2 x 20 mL). The organic layer was dried over MgSO4, filtered and evaporated to afford the crude
product. The crude was purified by flash silica chromatography with a 20-70% elution gradient of 3:1
EtOAc: EtOH in heptane. The title compound was obtained as a pale yellow solid after evaporation of
the solvents (69.2 mg, 74.4 %). 1H NMR (400 MHz, CDCl3) 1.04 (3H, t), 1.40 (3H, t), 1.90 (2H, ddt),
1.99 – 2.11 (3H, m), 2.41 (2H, td), 2.83 (2H, t), 3.15 (2H, q), 5.25 (2H, s), 5.27 – 5.33 (1H, m), 5.74 (1H,
d), 7.55 (1H, dd), 7.64 (1H, s), 7.83 (1H, d), 8.23 (1H, d).
2-(4-(4-oxo-4-SN38 butyl)-1H-1,2,3-triazol-1-yl)acetic acid (1d). The reaction for 2c was repeated
starting with 300mg of SN-38 to yield 294mg (0.60 mmol) of product (79.0%). This product was
dissolved in anhydrous DMF (10 mL) in a nitrogen purged reaction vial. 2-azidoacetic acid (227 μL,
3.02 mmol) was added via Gilson pipette followed by copper(II) sulphate pentahydrate (15 mg, 0.06
mmol) and sodium ascorbate (24 mg, 0.12 mmol). After 10 minutes the reaction had gone to
completion as observed by LCMS and the reaction mixture was diluted with ethyl acetate (100 mL)
and was washed with 0.1M HCl (3 x 25 mL) and saturated brine (3 x 25 mL). The organic phase was
dried over magnesium sulphate and dried under vacuum. The crude was purified by flash silica
chromatography with an elution gradient of 10-80% 3:1 EtOAc: EtOH + 10% AcOH in heptane. The
title compound was obtained as a pale yellow solid after evaporation of the solvents (278 mg, 78 %).
1H NMR (400 MHz, d-DMSO) 0.90 (3H, q), 1.31 (3H, t), 1.89 (2H, tq), 2.06 (2H, p), 2.71 – 2.85 (4H, m),
3.20 (3H, q), 5.19 (2H, s), 5.35 (2H, s), 5.44 (2H, s), 6.48 (1H, s), 7.34 (1H, s), 7.68 (1H, dd), 7.92 (1H,
s), 8.03 (1H, d), 8.22 (1H, d).
G5-PLL[PMOx]x[1a]y (2a). G5 PLL-[PMOx]x[Azide]y (150 mg, 1.89 µmol) was dissolved in water (2
mL) in a nitrogen purged reaction vial. 1a (44 mg, 0.08 mmol) dissolved in methanol (6 ml) was
added followed by copper(II) sulphate pentahydrate (2.4 mg, 9.45 µmol) and sodium ascorbate (3.7
mg, 0.02 mmol). The reaction mixture was stirred for 1 hour before reducing the volume of the
10

ACCEPTED MANUSCRIPT
reaction mixture by half under vacuum. The product was purified by size exclusion column (LH20) in
methanol, collecting 0.5 mL fractions which were evaporated, taken up in deionised water and
freeze-dried. The polymer fractions were collated to give the title compound as a pale yellow solid
(184 mg, 96 %). D.L. = 8.1 wt. %. 1H NMR (400 MHz, D2O) 1.12 – 1.86 (br, 367H), 1.89 – 2.3 (br,
1500H), 3.19 (br, 136H), 3.56 (br, 1940H), 3.92 (br, 111H), 4.04 – 4.44 (br, 130H), 6.92-8.04 (br, 98H,
SN-38).
G5-PLL[PMOx]x[1b]y (2b). This product was synthesised as for 2a to yield a pale yellow solid (125
mg, 69 %). D.L. = 8.5 wt.%. 1H NMR (400 MHz, D2O) 1.12 – 1.86 (br, 367H), 1.89 – 2.3 (br, 1500H),
3.19 (br, 136H), 3.56 (br, 1940H), 3.92 (br, 111H), 4.04 – 4.44 (br, 130H), 6.92-8.04 (br, 101H, SN-38).
G5-PLL[PMOx]x[1c]y (2c). This product was synthesised as for 2a to yield a pale yellow solid (133 mg,
73 %). D.L. = 8.3 wt. %. 1H NMR (400 MHz, D2O) 1.12 – 1.86 (br, 367H), 1.89 – 2.3 (br, 1500H), 3.19
(br, 136H), 3.56 (br, 1940H), 3.92 (br, 111H), 4.04 – 4.44 (br, 130H), 6.92-8.04 (br, 103H, SN-38).
G5-PLL[PMOx]x[1b’]y. (2b’) Product 1b’ (250 mg, 0.43 mmol) was dissolved in anhydrous DMF
(10mL) and DMTMM.BF4 (225mg, 0.69mmol) was added. The reaction mixture was stirred for 10
minutes at room temperature. G5-PLL[PMOx]x[NH2]y (820 mg, 10.71 µmol) and triethylamine (48μL,
0.34mmol) dissolved in anhydrous DMF (5 mL) were added and the reaction mixture stirred under
nitrogen at room temperature for 5 hours (the reaction progress was regularly followed by LC-MS at
254 nm). The reaction mixture was diluted with ethyl acetate (15 mL) and then added dropwise to
rapidly stirring diethyl ether (500 mL) to form a pale yellow precipitate. The precipitate was collected
and dried under vacuum at room temperature for 1 hour. The solid was taken up in 50 mL of a 2:1:1
mixture of 0.1M KH2PO4 buffer (pH3), DI water and ethanol before purifying by VivaFlow (30 kDa
MWCO) for several cycles, before changing over to 100% deionised water when the solution was
absent of impurities and free drug as detected by LC-MS. The title compound was obtained as a pale
11

ACCEPTED MANUSCRIPT
yellow solid after freeze-drying (850 mg, 84 %). D.L. = 7.7 wt. %. 1H NMR (400 MHz, D2O) 1.12 – 1.86
(br, 367H), 1.89 – 2.3 (br, 1500H), 3.19 (br, 136H), 3.56 (br, 1940H), 3.92 (br, 111H), 4.04 – 4.44 (br,
130H), 6.92-8.04 (br, 94H, SN-38).
Compound formulation
For i.p. dosing, irinotecan was formulated in 7.5% DMSO + 92.5% water for injection (v/v); for i.v.
dosing, irinotecan was formulated in 2.5% DMSO + 97.5% water for injection (v/v) and sterile-
filtered. SN-38 was dissolved in saline according to the following method: SN-38 was dissolved in
NaOH solution (2M) (3.5 μL per 1 mg SN-38). The bright yellow solution was diluted with the
required volume of saline and then HCl (2M) (1 μL for every 1 mg of SN-38 used) was added to
ensure a stable, near-neutral solution was achieved. Dend-SN-38 and unloaded dendrimer were
hydrated in saline and sterile-filtered before injection.
Dend-SN38 free drug loading
An accurately weighed quantity of Dend-SN38 (1-2 mg) was weighed directly into a HPLC vial and
dissolved in 1 mL of a 2:1 Methanol: KH2PO4 (0.1 M) mixture. The samples were measured by HPLC-
UV immediately after dissolution (to minimise any release of SN-38) and the free drug content was
reported as a weight percentage from an average of 2 runs per sample.
Dend-SN38 total drug loading
Full release of SN-38 from the dendrimer was achieved by dissolving an accurate quantity of Dend-
SN38 (1-2 mg) in sodium hydroxide solution (0.2 M, 1 mL) in a sealed HPLC vial and leaving at room
temperature overnight (whilst prior hydrolysis studies having shown full release under these
conditions in approximately 2 h (data not shown)). The drug loading was measured by HPLC-UV (254
nm), 1% ammonia in place of TFA) against a 6 point calibration curve for the ring-open (carboxylate)
12

ACCEPTED MANUSCRIPT
form of SN-38 dissolved in sodium hydroxide solution (0.2 M). The total drug loading was reported
as a weight percentage from an average of 2 runs per sample.
Dend-SN38 in vitro release rates
Dend-SN38 conjugates (2a-c) were weighed into HPLC vials (2 vials of 1.5 mg per sample) and Wistar
rat plasma (500 μL) was added and gently mixed briefly until the conjugates had dissolved. Rat
plasma was utilised to ensure that a sufficient volume for all experiments could be obtained in a
single batch to eliminate batch-to-batch variation. The vials were sealed and incubated at 37°C. 25
μL aliquots were taken at time points 0.25 h, 0.5 h, 1 h, 2 h, 4 h, 8 h, 24 h, 48 h, 72 h and 96 h. The
samples were added directly to a Mini-UniPrep filter vial together with 300 μL 10% AcOH in ACN.
The sample was agitated for 2 minutes before filtering. 200μL of the filtered solution was added to a
separate HPLC vial with 300 μL of a 2:1 methanol: KH2PO4 (0.1 M) mixture and agitated before
immediately measuring the released SN-38 content by HPLC-UV (265 nm). Full release of the SN-38
was measured by removing 25 μL of the incubated plasma and adding 5μL of 2 M NaOH and then
incubating at 37°C overnight in a sealed 96-well plate. The plate was cooled over an ice bath and
then the samples were neutralised by addition of 5 μL of 2 M HCl and repeating the same steps as
above accounting for the difference in volume when calculating the released SN-38. Release
measurements in PBS were made similarly, except that 25 μL aliquots were added directly to a HPLC
vial with 300 μL of a 2:1 methanol: KH2PO4 (0.1 M) mixture and measured directly by HPLC-UV.
Animal models
All animal studies were conducted in accordance with UK Home Office legislation, the Animal
Scientific Procedures Act 1986, and the AstraZeneca Global Bioethics policy. All experimental work is
outlined in project license 40/3483, which has gone through the AstraZeneca Ethical Review Process.
Female nude (Hsd: Athymic Nude-Foxn1 nu) mice were purchased from Harlan Laboratories (now
Envigo; UK). All mice weighed more than 18 g at the time of the first procedure. For studies in
13

ACCEPTED MANUSCRIPT
SW620 tumour-bearing mice, 1 x 106 cells in 50% Matrigel were implanted subcutaneously onto the
left flank of mice. Tumour growth was monitored twice weekly using intersecting calliper
measurements; tumour volume was calculated from 3.14*length*width2/6000, where length is the
longer dimension. Bodyweights were recorded at the time of tumour measurement, and daily if
weight loss exceeded 8%. Mice were randomised based on tumour volume to ensure an equal
volume distribution across groups.
Dend-SN38 pharmacokinetics
To determine the plasma SN-38 concentrations achieved by the Dend-SN38 conjugates with
different release rates, naive female nude mice (n = 2/group) were injected intravenously with a
single dose of slow, medium or fast release Dend-SN38 at 4 mg/kg. Live bleeds were collected at
various time points post-dose for determination of concentrations of free and bound SN-38 via
UPLC-MS/MS (conditions in Supporting Information Tables S2 and S3) against an SN-38 calibration
curve (11 point standard calibration curve (1-10000 nM) prepared in DMSO and spiked into blank
plasma). Briefly, 20 µL of blood was collected from the tail vein using capillary tubes. Blood was
diluted with 80 µL of PBS and centrifuged at 13,000 rpm for 5 min at 4°C. The supernatant was
collected and stored at -80°C until analysis. At the time of analysis, the samples were defrosted over
ice and two separate aliquots of 30 μL were taken from each sample for liberated and total SN-38
concentrations. For liberated SN-38, 120 μL of methanol/acetonitrile (50/50) + 1% acetic acid was
added to the first aliquot to precipitate protein and stabilise any residual dendrimer-bound SN-38.
The precipitate was agitated for 45 seconds before being filtered using a Mini-UniPrep Syringeless
Filter (0.45 μm). 50 μL of the filtrate was transferred to a deep 96-well plate and diluted with
deionised water (300 μL) and frozen at -80°C prior to analysis. For the second aliquot where total SN-
38 levels were desired, the sample was placed into a 96-well plate and 5 μL of sodium hydroxide (2
M) was added. The plate was sealed with adhesive 96-well sealing tape and incubated at 40°C for 2
hours and was then placed on to an ice bath for 5 minutes before removing the sealing tape. 5 μL of
14

ACCEPTED MANUSCRIPT
HCl (2 M) was added and the well contents were transferred to a Mini-UniPrep Syringeless Filter
(0.45 μm) followed by 110 μL of acetonitrile/methanol (50/50) + 1% acetic acid and the same
procedure for filtration of protein and transfer to deep 96-well plate was followed as for the
liberated levels of SN-38. Whilst some of the intact dendrimer was removed by the acetonitrile
protein precipitation, additional steps to prevent post-workup release form any intact dendrimer
was minimised by addition of KH2PO4 buffer. Samples were measured immediately after workup or
frozen at -80oC until measurements could be made.
SN-38 pharmacokinetics
To characterise the plasma pharmacokinetics of SN-38 (carboxylate form), SW620 tumour-bearing
mice (n = 5) were injected intravenously with a single dose of SN-38 at 2 mg/kg. Live bleeds were
collected at various time points post-dose for determination of SN-38 concentrations via UPLC-
MS/MS. The same procedure for sample preparation was used as for the Dend-SN38
pharmacokinetics above.
Analysis of SN-38 concentration in tumours
Tumours were weighed and put into FastPrep Lysing Matrix A Tubes (MP Biomedicals UK). Three
times the volume of water per tumour weight was added to the tumour for homogenisation. The
samples were homogenised using a FastPrep-24™ machine (MP Biomedicals UK) at 6 m/s for 45
seconds. The cycle was repeated twice to gain complete homogenisation. Control tissue from
undosed animals was homogenised as above and spiked with an 11 point standard calibration curve
(1-10000 nM SN-38) prepared in DMSO. Tissue homogenates (25 µL) were precipitated with
acetonitrile (100 µl) containing internal standard, followed by centrifugation at 4500 rpm for 10
minutes. The supernatant (50 µl) was then diluted in 300 µl water and analysed via UPLC-MS/MS.
Effect of Dend-SN38 release rate on anti-tumour efficacy
15

ACCEPTED MANUSCRIPT
To investigate the effect of Dend-SN38 release rate on anti-tumour activity and therapeutic index, an
efficacy study was completed in mice bearing SW620 tumours. Animals were randomised for dosing
when the average tumour volume reached ~0.2 cm3. Mice (n = 8-10 mice/group) were dosed once
weekly with saline (i.v.), irinotecan at 25 mg/kg (i.v.), irinotecan at 50 mg/kg (i.p.), unloaded
dendrimer at 50 mg polymer/kg (i.v.), or slow, medium or fast release Dend-SN38 conjugates at 4
mg SN-38/kg (i.v.). A dose of 4 mg SN-38/kg required administration of 50 mg Dend-SN38/kg,
equalling ~8% drug loading. Unless specified otherwise, the reported dose for the dendrimer
conjugates is the dose of SN-38, not the polymer dose. Mice treated with irinotecan or medium
release Dend-SN38 received 4 doses, while all other groups received 3 doses before tumour burden
necessitated euthanasia as defined in the Project Licence. At 24 h after the final dose, mice were
euthanised and the gastrointestinal tract (n = 5/group) was excised, fixed in formalin, and embedded
in paraffin for assessment of changes to the gastrointestinal tract suggestive of drug toxicity.
Statistical significance was determined versus Control at day 16 post-dose using a one-sided two-
sample unequal variance t-test.
Effect of Dend-SN38 dose and schedule on anti-tumour efficacy
To explore how the dose and schedule of the medium release Dend-SN38 formulation influenced the
therapeutic activity, an efficacy study was completed in mice bearing SW620 tumours. Animals were
randomised for dosing when the average tumour volume reached ~0.2 cm3. Mice were dosed (n =
9/group) once weekly for 4 cycles with saline (i.v.), irinotecan at 50 mg/kg (i.p.), or medium release
Dend-SN38 at 2, 4, or 8 mg/kg (i.v.). At 24 h after the 4th dose, 3 mice per group were euthanised to
assess GI toxicity. The remaining mice were kept on study to evaluate tumour re-growth after the
final dose. The effect of dose scheduling was also investigated by administering the medium release
Dend-SN38 every 2 weeks for 2 cycles at 4 or 8 mg/kg (i.v.). Statistical significance was determined
versus Control at day 23 post-dose using a one-sided two-sample unequal variance t-test.
16

ACCEPTED MANUSCRIPT
Tumour pharmacodynamic effect of irinotecan and Dend-SN38
An acute dosing study was completed in SW620 tumour-bearing mice to compare the time course of
tumour DNA damage. To ensure sufficient tumour material was available for analysis, animals were
randomised for dosing when the average tumour volume reached ~0.5 cm3. Mice (n = 3-4/group)
were treated with a single dose of irinotecan at 50 mg/kg (i.p.) or medium release Dend-SN38 at 4
mg/kg (i.v.). At 24 h, 72 h, or 120 h post-dose, mice were euthanised and the tumour was excised
and fixed in formalin and embedded in paraffin.
Assessment of gastrointestinal toxicity
To compare the severity of gastrointestinal toxicity caused by treatment with irinotecan and the
Dend-SN38 conjugates, haematoxylin and eosin (H&E)-stained sections (4-6 µm) of formalin-fixed
paraffin-embedded (FFPE) duodenum, jejunum, ileum, and colon were examined by a trained
pathologist. Each segment was assessed and the severity of histopathological changes (crypt
degeneration and atrophy) was graded using a semi-quantitative (0-5) scoring system (0, none; 1,
minimal; 2, mild; 3, moderate; 4, marked; 5, severe).
Assessment of DNA damage in tumours
Immunohistochemical staining for ɣH2A.X was completed to assess DNA damage in tumours. FFPE
tumour sections (4 µm) were dewaxed with xylene and rehydrated through graded ethanols to
water. Antigen retrieval was performed in pH 9 buffer at 110°C for 2 min in a RHS histoprocessor
microwave. Endogenous enzyme activity was blocked with 3% hydrogen peroxide. Slides were
incubated with serum free protein block (Dako) for 20 min before the primary antibody (anti-
phospho-histone H2A.X (Ser139); 1:300; Cell Signalling Technology) was added for 1 h at room
temperature. The Dako Envision+ (anti-rabbit) peroxidase-labelled polymer system was applied for
30 min, and slides were then incubated with 3,3’-diaminobenzidine (DAB; Dako) for 10 min. Slides
were counterstained with Carazzi’s haematoxylin, dehydrated through graded ethanols to xylene,
17

ACCEPTED MANUSCRIPT
and coverslipped. Positive staining was measured using an in-house analysis developed using
ImageScope (v11.2.0.780; Aperio). Statistical significance between treatments was determined using
one-way ANOVA with Sidak’s multiple comparisons post-test.
Detection/localisation of SN-38 in tumour using Desorption Electrospray Ionization – Mass
Spectrometry Imaging (DESI-MSI)
Snap-frozen tumour tissues were cryo-sectioned using a Leica CM3050 S (Leica Microsystems)
cryostat to a thickness of 14 µm, thaw mounted onto Superfrost microscope slides and stored at -
80°C until analysis. DESI imaging experiments were performed on a Thermo Scientific Q-Exactive
instrument (Bremen, Germany) equipped with a 2D automated DESI source from Prosolia Inc.
(Indianapolis, IN, USA) using a home-built sprayer assembly as described previously [42]. Detection
of SN-38 was performed in negative ion mode as [M-H]-, while Irinotecan was detected in positive
ion mode as [M+H]+. To allow detection of both compounds in a single imaging experiment,
negative and positive ion mode were alternated during analysis. Analysis was performed at R =
70,000 mass resolving power and 50 V S-Lens Voltage. Methanol/water (95: 5 v/v) was used as
electrospray solvent at 4.5 kV spray voltage and a flow rate of 1.5 µL/min. Solvent was delivered
using a Dionex Ultimate 3000 stand-alone nanoLC pump (Thermo Scientific). Nitrogen N4.8 was used
as nebulising gas at a pressure of 7 bar. Distance between sprayer to sample surface was 1.5 mm
while distance between sprayer and MS inlet was 7 mm. The spray angle was set to 75° while
collection angle was set at 10°. The capillary temperature was set to 320°C. Data was converted into
centroid .imzML format using imzML converter version 1.1.4.5 [43] and visualised using MSiReader
v0.05 [44]. All intensities shown are raw ion intensities. 1st order linear interpolation was used for
image generation. Samples were stained with H&E post-DESI analysis and digitalised using a Leica
Aperio AT2 slide scanner (Leica Microsystems).
Measurements to determine the sensitivity of the DESI-MSI technique
18

ACCEPTED MANUSCRIPT
SN-38 and Irinotecan standard solutions dissolved in 50% MeOH (diluted from a stock solution in
DMSO) were spotted onto six adjacent tumour tissue sections. 1 μL of standard solutions with
concentrations ranging from 50 uM to 0.1 uM was spotted taking care to ensure that the standards
were spotted onto comparable positions of the tissue sections. Tissue sections were subsequently
analysed by DESI-MSI using the experimental parameters as described for the SN-38 localisation.
To calculate on-tissue concentrations, the assumption was made that standard solutions had
distributed evenly throughout the tissue volume under the calibration spot. Tissue thickness was 10
μm for all sections and the area of the calibration spot was calculated using the number of pixels
within a calibration spot with spatial resolution used in the experiment. Concentrations within the
calibration spots ranged from 1146.8 μM to 2.4 μM. Signal to noise (S/N) for the lowest
concentration spot was determined to be on average 7.2 and 6.7 for SN-38 and Irinotecan
respectively, thus lying below the limit of quantification at S/N=10. For several pixels of the lowest
concentration spot S/N fell below 3, indicating the signal was below the limit of detection.
3. Results & Discussion
Design and synthesis of SN-38 linkers and dendrimer conjugation
After synthesising a dendrimer platform with the desired size, surface-modification, and
biocompatibility [34], we then developed SN-38 linkers with a range of release rates to explore
differential SN-38 profiles over time. Three linker chemistries were utilised to tune the release rate:
a primary amine carbamate, an ester, and a secondary amine carbamate, respectively (Scheme 1).
To generate fast and medium release linkers, the primary amine carbamate and ester linkers were
synthesised to have sufficient distance from the amide of the linker itself (carbamate) or from the
amide from the dendrimer coupling (carbamate and ester) to avoid anchimeric assistance to
hydrolytic breakdown. This effect has been shown to have a significant impact on release rates [45],
and was exploited to increase release rate. Inclusion of a triazole in the linker design as
19

ACCEPTED MANUSCRIPT
demonstrated here would be expected to reduce this effect by extending and rigidifying the linker
such that any transition state involving anchimeric assistance would be highly strained and
energetically disfavoured. To generate a fast release linker, the 1° amine carbamate was used for its
increased susceptibility to chemical hydrolysis, which we hypothesise is due to a mechanism
involving the formation of an intermediate isocyanate. This isocyanate decomposes rapidly in the
presence of water to give a primary amine and carbon dioxide by-product. Hence, to produce a slow
release linker, we investigated the N-methylated 2° amine carbamate, which cannot undergo this
type of release mechanism. Consistent with our hypothesis, this linker exhibited a slower release
rate. This exemplifies how the dendrimer platform can be coupled with unique linker chemistries to
achieve different release rates to provide desired drug profiles.
20

ACCEPTED MANUSCRIPT
Scheme 1. Synthesis routes for linker-SN38 compounds 1a-c. (a) ACN/THF, DMAP, RT. (b) DCM,
EDC/DMAP RT, (c + d) DCM, DIPEA, (e) DMF, sodium ascorbate, copper II sulphate, RT.
Initially, linkers 1a-c were designed with alkyne groups to conjugate to an azide-modified dendrimer
through click chemistry. The methodology for the dendrimer synthesis and linker-SN38 conjugation
are outlined in Scheme 2. At a later stage, an alternative route of synthesis was made for linker 1b
(linker 1b’) to avoid the use of copper click chemistry in the presence of the POZ-modified
dendrimer. POZ has been reported to have an affinity for copper in click reactions [46] which,
although not observed here, was still avoidable by carrying out the purification on the linker before
conjugating to the dendrimer.
1a-c
Copper
“Click”
G5 L-lysine
dendrimer
=
=
1b’
Amide
coupling
Scheme 2. Synthetic route to POZ-modified dendrimers with linker-SN38 conjugation.
21

ACCEPTED MANUSCRIPT
The drug loading was consistent by employing click chemistry to conjugate SN-38 linkers 1a-c to the
dendrimer (Table 1). Unreacted SN-38 linker was efficiently removed from the conjugate using a LH-
20 sephadex size exclusion column with methanol eluent; unreacted linker was not detectable by
HPLC-UV (LoD ~ 0.1 μg/mL). Presumably, the unbound SN-38 detected in sample 2a arose from the
release of some of the drug during the evaporation and solvent changeover process, since this
conjugate had the fastest release rate. The unbound SN-38 in 2b’ was a small amount of residual
drug leftover from the purification process.
Conjugate 2b’ was synthesised differently: the linker was conjugated to the dendrimer through an
amide coupling step and was purified by ultrafiltration afterwards. This was the most efficient
method for the preparation of the conjugate since it removed an ultrafiltration purification step
from the dendrimer after the azide modification and was found to be a better method for larger
quantities of material. All of the dendrimer conjugates had ca. 8 wt.% of SN-38, which equates to 25
drug molecules per dendrimer on average. Residual functional groups were not quantified but are
expected to be minimal given the excess of drug-linker used and the consistencies in the loadings.
The solubility of the conjugates remained very good but reduced in comparison to the native
dendrimer. The structural representation of each dendrimer-SN38 conjugate (Dend-SN38) can be
found in Supporting Information Scheme S1. A representative GPC chromatogram and DLS spectra
have also been included for conjugate 2b’ in the Supporting Information Fig. S1.
Table 1. SN-38 loading and non-conjugated (free) SN-38 for the different dendrimer conjugates.
Dendrimer description
Release
Conjugated SN-38
wt.%
Free SN-38 in
sample wt.%
description
22

ACCEPTED MANUSCRIPT
2a (1° amine carbamate)
2b (Ester)¥
Fast
8.1
8.3
0.2
ND
Medium
2b’ (Ester)Ϯ
Medium
Slow
7.7
8.5
0.17
ND
2c (2° amine carbamate)
ND = Not Detected (by HPLC-UV). ¥ Conjugated by click chemistry. Ϯ Conjugated through amide coupling.
Dend-SN38 release rates
The release rates for SN-38 from the Dend-SN38 conjugates were measured in PBS, and
subsequently in plasma to mimic the in vivo situation more closely. The release rates measured for
the Dend-SN38 conjugates 2a-c in plasma (Fig. 1B) in vitro were found to differ from those measured
in PBS (Fig. 1A), presumably due to enzymatically mediated processes, despite the presence of the
polyoxazoline on the surface of the dendrimer. The range of in vitro plasma release half-lives for the
three Dend-SN38 conjugates was determined to be 2.5 h (“fast release”), 21 h (“medium release”),
and 72 h (“slow release”) (Fig. 1B). Based on the drug loading and a 21 h release rate, it was
predicted that the medium release Dend-SN38 would achieve the desired sustained exposure to
therapeutic concentrations of SN-38. The fast and slow release linkers were designed to give greater
insight into the relationship between efficacy and toxicity by providing exposure profiles outside of
the target plasma level.
Plasma SN-38 pharmacokinetics studies were performed in mice to confirm whether there would be
a prolonged, therapeutic level of SN-38 in the circulation in vivo (Fig. 1C). The plasma half-life of SN-
38 (carboxylate form) itself in vivo is less than 0.5 h (Supporting Information Fig. S2) and previous
reports have shown that the SN-38 generated after administration of irinotecan is rapidly cleared
[47, 48]. Of the three Dend-SN38 conjugates we designed, only the medium release conjugate
achieved the desired sustained exposure of SN-38 in vivo. Therapeutic concentrations of SN-38 were
measured up to 96 h after the administration of this conjugate. As expected for the fast release
23

ACCEPTED MANUSCRIPT
conjugate, the plasma levels of SN-38 quickly diminished within 10-20 h after administration. This is
consistent with the rapid release of SN-38 from the conjugate (t1/2 =2.5 h) and subsequent rapid
clearance of liberated SN-38. The slow release conjugate was designed to provide a low but
consistent exposure to SN-38. However, a short burst release was observed, followed by an
equilibrated concentration of SN-38, which was below the limit of detection of the instrument
(UPLC-MS/MS LoD = 5 nM). Given that there was no free drug detected in the sample by HPLC-UV,
the rationale for the early burst may have originated from some enzymatic action on the more easily
accessible linkers, since enzymatic cleavage of secondary amine carbamates is known [49]. It is also
possible that demethylation of enzyme-accessible carbamate groups could have occurred and would
also have led to faster release of SN-38. Some ex vivo release during sample storage and preparation
may have been a minor contributing factor. Interestingly, the total SN-38 plasma profile (i.e.
conjugated plus free SN-38) showed that the slow release conjugate was being cleared faster than
the medium release conjugate, despite the slower release rate. This is thought to be driven by the
residual functional groups exposed by the linkers after drug release, i.e. a carboxyl group for the
ester (medium release) versus a 2o amine for the carbamate (slow release), which could be
recognised by the mononuclear phagocyte system and cleared from circulation at different rates.
24

ACCEPTED MANUSCRIPT
In vitro SN-38 Release (Plasma)
B.
A.
In vitro SN-38 Release (PBS)
120
100
80
60
40
20
0
120
100
80
60
40
20
0
0
10
20
30
40
50
60
70
0
10 20 30 40 50 60 70 80
Time (h)
Time (h)
C.
100000
10000
1000
100
D.
Plasma Free SN-38 Concentration (In vivo)
Plasma Total SN-38 Concentration (In vivo)
100000
10000
1000
100
10
10
1
0
20
40
60
Time (h)
80
100
0
20
40
60
80
100
Time (h)
Ester
(Medium)
2° amine carbamate
(Slow)
1° amine carbamate
(Fast)
Figure 1. (A) In vitro release of SN-38 in PBS. (B) In vitro plasma release (Wistar rat). (C) In vivo
pharmacokinetics of SN-38 liberated from the Dend-SN38 conjugates (2a-c). (D) In vivo
pharmacokinetics for total SN-38 (liberated + conjugated).
In vivo efficacy and safety
The relative anti-tumour efficacy of the Dend-SN38 conjugates at 4 mg/kg versus irinotecan was
determined using mice bearing subcutaneous SW620 colorectal xenografted tumours. This model
was selected for its well-characterised response to irinotecan. Dend-SN38 conjugates were
administered weekly intravenously, at doses expected to achieve therapeutic concentrations of SN-
38 for different periods of time, based on in vitro and in vivo release and pharmacokinetic studies. By
mass, irinotecan is ~67% SN-38. Irinotecan was dosed at the maximum tolerated i.v. and i.p. doses to
test two different profiles of plasma exposure and local drug concentration, with the SN-38
concentrations related proportionally. Intravenous administration achieves
a
high initial
25

ACCEPTED MANUSCRIPT
concentration, followed by rapid clearance; intraperitoneally dosed irinotecan reaches the plasma
more slowly and achieves more prolonged drug exposure with a lower Cmax. This more closely mimics
the clinical standard-of-care of irinotecan infusion.
The SW620 xenograft model has a well characterised, but modest, response to irinotecan. Both i.p.
and i.v. irinotecan treatment caused a slight reduction in tumour growth (50% inhibition, P<0.01 and
19% inhibition, P>0.05, respectively), but the tumours continued to progress (Fig. 2A). Changes in
the growth rate of i.p. irinotecan-treated tumours were observed by day 3 post-dose, whilst little
response was observed for the i.v. treatment. This supports the value of more prolonged input of
drug into plasma following i.p. dosing. In contrast to the modest anti-tumour effect of irinotecan,
treatment with the medium release Dend-SN38 led to significant tumour regression (P<0.001; Fig.
2A). The deviation from saline controls occurred on day 5 post-dose (versus day 3 post-dose for
irinotecan) and tumour regression was evident by the third cycle of dosing. At the end of study,
medium Dend-SN38 treatment had caused 69% tumour regression, while regression was not
observed with irinotecan treatment. Highlighting the relationship between drug exposure profile
and anti-tumour efficacy, the fast and slow release conjugates gave no anti-tumour effect (Fig. 2A).
This is consistent with the low plasma concentrations of released SN-38 seen in the in vivo
pharmacokinetic evaluation (Fig. 1C). The sustained therapeutic concentration of SN-38 delivered by
the medium release conjugate resulted in more than a 3-fold reduction in average tumour volume at
the end of the study. It will be important to test the robustness of the Dend-SN38 conjugate across a
panel of pre-clinical models to develop greater insight into its clinical potential.
26

ACCEPTED MANUSCRIPT
B.
A.
B o d y w e ig h t
% C h a n g e
E ffic a c y in S W 6 2 0 X e n o g ra ft in M ic e
1 0
5
2 .0
**
**
S a lin e i.v . q 7 d x 3
S a lin e
Irin o tec a n 25 m g /k g i.v . q 7 d x4
Irin o tec an 50 m g /k g i.p . q 7d x4
1 .0
Irin o te c a n i.p .
Irin o te c a n i.v .
F a s t D e n d -S N 3 8
M e d iu m D e n d -S N 3 8
S lo w D e n d -S N 3 8
E m p ty D e n d rim e r
0 .5
**
F a s t D e n d -S N 3 8
M e d iu m D e n d -S N 3 8
S lo w D e n d -S N 3 8 m g /k g i.v . q 7d x3
C o n tro l D e n d rim e r 5 0 m g p o ly m e r/kg i.v . q 7 d x3
4
m g /k g i.v . q 7d x3
4
m g /k g i.v . q 7 d x 4
0 .2 5
4
0
***
0 .1
-5
0 .0 5
0
5
1 0
1 5
2 0
2 5
0
5
1 0
1 5
2 0
2 5
D a y s o f D o s in g
D a y s o f D o s in g
D.
C.
ii
i
iii
iv
Figure 2. Efficacy and gastrointestinal toxicity of Dend-SN38 conjugates (2a-c) versus irinotecan
both i.p. and i.v. at the maximum tolerated dose. (A) Mouse SW620 anti-tumour efficacy plot
dosing q7d. (B) Mean group bodyweight change during efficacy study. (C) Gastrointestinal toxicity
score for crypt atrophy; 0 = no change, 5 = severe changes. (D) Representative gastrointestinal
tract cross sections of crypt epithelium at 24 h post-last dose; (i) Control, (ii) Irinotecan i.v., (iii)
Irinotecan i.p., (iv) Medium release Dend-SN38. ** and *** = P< 0.01 and P<0.001 versus Saline at
Day 16, respectively.
Pre-clinically and clinically, treatment with irinotecan can be limited by certain toxicities, including
gastrointestinal toxicity [50]. Gastrointestinal toxicity resulting from irinotecan treatment is thought
to be driven by a number of mechanisms, including direct conversion of irinotecan into SN-38 in the
intestines, and enzymatic cleavage of the metabolite SN-38G (SN-38 glucuronide) back into SN-38
during enterohepatic circulation and reabsorption of the drug [51]. These toxicities can be observed
pathologically in mice. To investigate the effect of repeated doses of the different treatments on the
gastrointestinal tract, the toxicity was assessed at the end of the study. A pathologist-based
assessment of H&E-stained samples of the gastrointestinal tract was performed (see Fig. 2D for
27

ACCEPTED MANUSCRIPT
representative images). Baseline toxicity profiles were established using i.v. and i.p. irinotecan
treatments as positive controls. Irinotecan treatment resulted in minimal to mild degenerative
effects within the small intestinal crypt epithelium, with generally more severe concomitant atrophic
changes with i.p. irinotecan treatment (Fig. 2C). This most likely reflects the greater local
gastrointestinal exposure [52] and extended plasma exposure [53] to irinotecan/SN-38 with the i.p.
dosing. The pathological analysis of crypt degeneration correlates with mean mouse bodyweight loss
for the different treatment groups: irinotecan i.p. caused the greatest bodyweight loss, with smaller
weight fluctuations for irinotecan i.v. and medium release dendrimer conjugate, and good
tolerability for other treatments (Fig. 2B).
The toxicity of the dendrimer conjugates was compared directly to the baseline irinotecan toxicity
profiles. Each conjugate presented a unique profile of toxicity. While the slow release dendrimer
conjugate had no impact on the gastrointestinal tract, it also did not exert any therapeutic effect.
The findings suggest that the concentration of SN-38 in the tumour was too low to provide efficacy.
Similarly, the fast release dendrimer conjugate showed no anti-tumour activity. However, this is
thought to be due to the rapid release of the drug and subsequent rapid clearance of the free drug,
therefore, not providing a prolonged exposure of the drug required in the tumour for efficacy.
Moreover, this conjugate caused minimal to mild toxicity throughout the small intestine (similar to
i.v. irinotecan treatment), which is thought to be due to higher initial plasma concentrations
resulting from more rapid drug release. This suggests that the plasma drug exposure is critical to
managing the toxicity of irinotecan/SN-38. In contrast, the significantly improved efficacy of the
medium release Dend-SN38 conjugate was accompanied by reduced gastrointestinal effects relative
to both irinotecan groups and the fast release Dend-SN38. No atrophic effects were observed in the
ileum or duodenum, though a small number of animals showed minimal to mild crypt atrophy in the
jejunum after treatment with the medium release dendrimer conjugate. These observations are
consistent with a reduction in the plasma Cmax of SN-38 when administered as a polymer therapeutic
28

ACCEPTED MANUSCRIPT
with a prolonged release rate, thus a reduced distribution to the liver and intestines leading to lower
toxicity.
The combination of potent tumour regression and good gastrointestinal tolerability with the
medium release dendrimer conjugate is a significant improvement in the therapeutic index of SN-38.
The superiority of the medium release Dend-SN38 over the slow and fast release conjugates made it
the logical formulation to investigate further. Additional studies were completed to explore the
efficacy and pharmacodynamic effects of the medium release conjugate. Henceforth the medium
release Dend-SN38 conjugate will be referred to as DEND-38.
B.
A.
(i)
S W 6 2 0 X e n o g ra ft in M ic e
2
1
S a lin e i.v . q 7 d
(ii)
Irin o tec a n 5 0 m g /kg i.p . q 7d
0.5
0.25
**
D E N D -3 8
D E N D -3 8
D E N D -3 8
2
4
8
m g /k g i.v . q 7d
m g /k g i.v . q 7d
m g /k g i.v . q 7d
0.1
***
&
&
0.05
(iii)
(iv)
***
***
0.01
&
=
1
m o u s e w ith o u t v is ib le
tu m o u r at en d o f s tu d y
0.001
0
7
1 4 2 1 2 8 3 5 4 2 4 9 5 6 6 3 7 0 7 7 8 4 9 1 9 8
D a y s o f D o s in g
(v)
S a lin e i.v . q 7 d
5 0 g /k g Irin o te c a n i.p . q 7 d
S a lin e i.v . q 7 d
C.
m
D. (i)
2 m g /k g D E N D -3 8 i.v . q 7 d
(ii)1 2 0
1 2 0
1 1 0
1 0 0
9 0
S W 6 2 0 X e n o g ra ft in M ic e
1 1 0
1 0 0
9 0
2
S a lin e i.v . q 7 d
Irin o tec a n 5 0 m g /kg i.p . q 7d
1
0 .5
**
D E N D -3 8
D E N D -3 8
D E N D -3 8
D E N D -3 8
4
8
4
8
m g /kg i.v . q 1 4d
m g /kg i.v . q 1 4d
m g /k g i.v . q 7d
m g /k g i.v . q 7d
8 0
**
8 0
0 .2 5
0
5
1 0
1 5
2 0
2 5
0
5
1 0
1 5
2 0
2 5
D a y s o f D o s in g
***
D a y s o f D o s in g
0 .1
S a lin e i.v . q 7 d
S a lin e i.v . q 7 d
g /k g D E N D -3 8 i.v . q 7 d
***
***
0 .0 5
8
m
4
m g /k g D E N D -3 8 i.v . q 7 d
(iii) 1 2 0
1 2 0
(iv)
1 1 0
1 0 0
9 0
1 1 0
1 0 0
9 0
0 .0 1
0
7
1 4
2 1
2 8
3 5
D a y s o f D o s in g
8 0
8 0
0
5
1 0
1 5
2 0
2 5
0
5
1 0
1 5
2 0
2 5
D a y s o f D o s in g
D a y s o f D o s in g
Figure 3. (A) Dose-dependent efficacy for DEND-38 and 72 day re-growth period following final
dose. (B) Picture of tumours excised at 24 h post-last dose (day 23) representing treatment with (i)
saline, (ii) irinotecan i.p., (iii) DEND-38 2 mg/kg, (iv) DEND-38 4 mg/kg, and (v) DEND-38 8 mg/kg.
29