Preclinical evaluation of a next-generation, EGFR targeting ADC
that promotes regression in KRAS or BRAF mutant tumors
AACR 2016
Abstract #
1217
Lei Huang1, Bob Veneziale1, Mark Frigerio2, George Badescu2, Xiaoming Li1, Qiping Zhao1, Jesse Bahn1, Jennifer Souratha1, Ryan Osgood1, Chunmei Zhao1,Kim Phan1, Jessica Cowell1, Sanna Rosengren1, Anas Fathallah1, Jason Parise1, Martin Pabst2, Mathew Bird2,
William McDowell2, Gina Wei1, Curtis Thompson1, Antony Godwin2, H. Michael Shepard1, and Christopher Thanos1
1Halozyme Therapeutics, San Diego, CA, USA; 2Abzena, Antitope Limited & PolyTherics Limited, Cambridge, United Kingdom
ABSTRACT
RESULTS
RESULTS (cont.)
RESULTS (cont.)
•
•
ThioBridge™ ADC conjugation was performed as described in Ref 10. MMAE corresponds to monomethyl auristatin E.
•
Tumor xenograft growth studies were conducted in Ncr nu/nu mice. MDA-MB-231 cells were implanted in the mammary fat
pad, and HT-29 (colorectal adenocarcinoma) cells were injected subcutaneously, both at 3.3 x 106 cells/site.
•
•
•
•
Mutation promotes cell
growth and is resistant to
mAb therapy
ADC overcomes
mutation resistance and
selectively kills tumor cell
Naked
mAb
EGF
EGFR
20

30
40
50
60
70
80
90
3
Summary
500

100
0
10
20
30
40
50
60
70
80
90
100 110
T im e ( D a y s )
Dosing stopped at Day 38
Dosing stopped at Day 39
Fig. 6 Targeting EGFR+, mutant tumors with an ADC. Halo Variant 1-MMAE conjugates (maleimide based
conjugation, vc-PAB-MMAE) were used to dose EGFR+, mutant tumors (bi-weekly) at either 1, 3, 10, or 30 mg/kg.
Cetuximab was dosed bi-weekly at 30 mg/kg. A dose response for Halo Variant 1-MMAE was observed in both
models. In both models, no activity was observed for cetuximab. N=6 mice group.
1
2
3
4
5
6
7
0
1
2
Random Lysine Conjugation
Limited dermal scoring findings comparable with vehicle control group
•
No unexpected findings observed at either dose (2.5 mg/kg and 8 mg/kg)
•
Safety profile met criteria for candidate nomination
HTI-1511 In Vivo Stability in Primates
HALO mAb Maleimide MMAE
HTI-1511
ER: 91%
3
4
5
25.0
6
7
8 9
0
1
2
3
4
5
6
7
20.0
5.0
12.4 min (1.8 %)
-2.0
7.5
Exposure Ratio (ER) = Group Mean AUC of ADC/Group Mean AUC of Total x 100
HTI-1511 Potency in EGFR+, KRASMUT PDx Models
NSCLC (EGFR+, KRASpG12C) PDx
Fig. 8 Generation of HTI-1511.
The HALO anti-EGFR mAb (see
Methods section) generated an
ADC through use of ThioBridge
conjugation technology, and
linked with a branched PEG
spacer to a val-cit PAB-MMAE
warhead, resulting in isolation of
98.2% DAR 4 species
preparation, as assessed by
analytical HIC.
MMAE
PEG side chain
10.0
5.0
Fig. 10 PK and in vivo ADC stability assessment.
HTI-1511 was more stable than a maleimide-based
ADC control. Dose=8 mg/kg for each ADC. n=3 for
HTI-1511, n=2 for HALO mAb Maleimide MMAE.
ThioBridge Chemistry
ThioBridge
HTI-1511
DAR 4 Peak
2.5
Toxin on ADC Detection
8 9
13.6 min (98.2 %)
15.0
10.0 12.5 15.0 17.5 20.0 22.5
KRASMut
KRASMut
BRAF
BRAF
BRAF
BRAF
MEK
MEK
MEK
MEK
ERK/MAPK
ERK/MAPK
ERK/MAPK
ERK/MAPK
Internalized ADC
BRAF
Migration
Angiogenesis
Survival
1000
last dose
500
0
20
40
60
10 mg/kg
2500
V e h ic le
H T I-1 5 1 1
2000
1500
1000
500
last dose
0
0
80
20
40
60
S tu d y D a y
S tu d y D a y
F c g R IIb
F c g R IIa
F c g R IIIa 1 5 8 V
10000
50000
H A LO m A b N aked
40000
H A L O m A b P T 2 -M M A E
20000
H A LO m A b N aked
H A LO m A b N aked
H A L O m A b P T 2 -M M A E
30000
6000
20000
4000
10000
2000
0
2
3
4
1
L O G ( u g /m L )
40000
10000
0
0
2
3
-1
4
0
L O G ( u g /m L )
F c g R IIIa 1 5 8 F
15 mg/kg
H A L O m A b P T 2 -M M A E
8000
H A LO m A b N aked
H A L O m A b P T 2 -M M A E
HA L
O
30000
m Ab N aked
3
4
H A LO m A b N aked
H A L O m A b P T 2 -M M A E
40000
2
FcgR I
FcRn
50000
1
L O G ( u g /m L )
H A L O m A b P T 2 -M M A E
20000
Vehicle
V
e h ic le
2500
Human TNBC
KRASG13D
MDA-MB-231M
HALO
mAb
H
ALO m
A b v cvcPAB-MMAE
P A B -M M A E
2000
V e h ic le
Vehicle
2500
HTI-1511
H
A L O m A b P T 2 -M M A E
2500
Vehicle
V e h ic le
2000
H A L OmAb
m A b vcPAB-MMAE
v c P A B -M M A E
HALO
HTI-1511
H A L O m A b P T 2 -M M A E
H ALO m
A b v cvcPAB-MMAE
P A B -M M A E
HALO
mAb
2000
HALO m Ab
HTI-1511
P T 2 -M M A E
1500
1500
1000
1000
1000
500
500
500
0
0
0
20
40
60
80
0
20
40
S tu d y D a y
60
20000
80
100
30000
20000
10000
10000
10000
0
0
1500
Fig. 11. KRAS-mutated PDx
tumor model studies. HTI-1511
was tested in EGFR+, KRAS
mutated, NSCLC or
cholangiocarcinoma models
(Champions Oncology). Dosing
was 2.5 mg/kg weekly, N=8 mice
per group. Tumor regressions
were observed for all mice. No
impact on body weight was
observed.
Attenuated Binding of HTI-1511 to FcRg Subtypes
30000
0
1
2
3
4
0
-2
0
2
4
L O G ( u g /m L )
L O G ( u g /m L )
-3
-2
-1
0
1
L O G ( u g /m L )
2
Fig. 12. HTI-1511 exhibited attenuated
binding to several FcRg subtypes when
compared to the parental mAb.
Binding was assessed by FACS. HTI-1511
demonstrated significantly attenuated
binding to FcγRIIa, FcγIIb, FcγIIIa 158V,
and FcγIIIa 158F receptors, but not
attenuated binding to FcγR1, suggesting
that HTI-1511 might have improved
tolerability due to lack of binding by
FcγRII-III receptors, possibly due steric
hindrance from the PEG side chain.
HTI-1511 PROGRAM SUMMARY
• A mAb was engineered with attenuated binding to human skin, compared to cetuximab
0
20
40
S tu d y D a y
60
80
100
• An ADC mechanism of action that delivers a cytotoxic payload can treat KRAS- or BRAF-mutated tumors in mice
S tu d y D a y
• HTI-1511 utilizes next generation, ThioBridge chemistry
V e h ic le
Vehicle
Human CRC
BRAFV600E
HT29
Vehicle
V e h ic le
3000
H ALO m
A b v cvcPAB-MMAE
P A B -M M A E
HALO
mAb
H
A L O m A b P T 2 -M M A E
HTI-1511
2000
2000
1000
1000
Vehicle
V e h ic le
3000
H ALO m
A b v cvcPAB-MMAE
P A B -M M A E
HALO
mAb
HALO
H A L O mAb
m A b vvcPAB-MMAE
c P A B -M M A E
HALO m Ab
HTI-1511
HALO m Ab
HTI-1511
P T 2 -M M A E
2000
P T 2 -M M A E
•
More homogeneous, active and stable (in primates) than maleimide based conjugation, in preclinical studies
• Safety profile in a pilot primate toxicology study met criteria for candidate nomination
• Tumor regressions observed in PDx tumor models that resulted in no remaining measurable tumors in mice
1000
• IND enabling studies underway
0
0
10
20
30
S tu d y D a y
Fig. 5 Can the cytotoxic mechanism of action of an anti-EGFR ADC overcome downstream, activating
mutations that are resistant to naked antibody therapy?
1500
0
REFERENCES
0
Released Cytotoxins
Tumor Cell Death
H T I-1 5 1 1
2000
1
0
Proliferation
V e h ic le
-1 0 0 0 0
5 mg/kg
Cholangiocarcinoma (EGFR+,KRASpG12A) PDx
2500
0
ADC
KRAS
ADC Detection
mAb Detection
Fig. 7 Testing alternative
conjugation chemistry.
ThioBridge conjugation
Technology10 leads to a
consistent drug:antibody
ratio (DAR) of 4, whereas
first generation chemistries
that are based on random
lysine or cysteine
conjugation lead to more
heterogeneous mixtures.9
# drugs / Ab
Analytical Hydrophobic Interaction
Chromatography (HIC)
mAU
ER: 40%
ThioBridge Cysteine Conjugation9
Cysteine Conjugation
3000
No signaling
•
More Homogeneous
# drugs / Ab
8 9
4
29
Dose
Ligand
Proliferation
3
22
1000
0
T im e ( D a y s )
0
KRAS
Migration
Angiogenesis
Survival
T u m o r V o lu m e (m m ) ± S E M
3
10
2
15
3
T u m o r V o lu m e (m m )  S E M
0
ADC = antibody-drug conjugate
Pilot Primate Toxicology. Experiments were performed at a CRO under veterinary supervision and appropriate ACUC
guidelines. Female animals were intravenously administered vehicle (phosphate buffered saline) or test article on Days 1 and
22. Animals were administered doses of 2.5 and 8.0 mg/kg (3 per group).
Fc Receptor binding studies. CHO cells were engineered to express various Fc receptors. The unconjugated antibody or HTI1511 was incubated with each of these cell lines, then washed with PBS, labeled with R-PE-conjugated F(ab’)2 fragment of
goat anti-human IgG (Jackson ImmunoResearch, Cat# 109-116-097, washed, then assessed by FACS.
Naked anti-EGFR mAb
inhibits signaling
pathway
1500
1
8
0
1
M FI
Day 3
Day 2
Fig. 4 In vivo imaging of
tumor vs. skin binding.
A human donor foreskin
xenograft and human
derived A431 tumor
xenograft were paired
and imaged over 7 days
following DyLight755
labeled cetuximab and
HALO mAb V1
administration. For HALO
mAb V1, the ratio of
binding between tumors
and skin was higher than
that of cetuximab
(P<0.05).
Rationale for Treatment of EGFR+ Mutation
Resistant Tumors
PDX Studies. Female Harlan nu/nu athymic nudes mice were implanted bilaterally with approximately 5x5x5 mm tumor
fragments subcutaneously in the left and right flanks with Champions TumorGraft™ models CTG-0828 and CTG-0941. When
tumors reached 1 to 1.5 cm3, they were harvested and viable tumor fragments approximately 5x5x5 mm were implanted
subcutaneously in the left flank of female study mice.
Primate Pharmacokinetics. An ELISA-based assay was used to measure antibody or ADC concentrations in cyno plasma
samples drawn under study protocol from the toxicology study a CRO (antibodies used were goat-anti-human IgG polyclonal
antibody from Bethyl Laboratories, Cat. No. A80-319A and mouse-anti-MMAE monoclonal antibody (Epitope Diagnostics, No.
MAB30699).
0
# drugs / Ab
HALO
Anti-EGFR
mAb
EGFR-mediated
signaling promotes cell
growth
500
(S E M )
•
1000
1st Gen Chemistries: Heterogeneous
METHODS
Recombinant antibodies were expressed as full-length IgG1 in CHO cells and purified with Protein A affinity resin. HALO antiEGFR mAb Variant 1 (V1) corresponds to an early antibody lead. The anti-EGFR mAb used in HTI-1511 conjugate is a further
engineered, humanized, candidate mAb.
In vitro TME-specificity ELISA. An EGFR-coated plate was subjected to increasing concentrations of either HALO mAb or
cetuximab in 25% human serum and 16.7 mM lactic acid for 60 minutes at either pH 6.0, pH 6.5, or pH 7.4 and washed with
PBS. A secondary antibody-HRP conjugate was then added, washed and detected at OD450.
For in vivo antibody binding studies, human foreskin skin grafts were implanted into Ncr nu/nu mice. Antibodies were labeled
with the near-IR fluoroprobe DyLight755NHS ester dye, and administered IV at 10 mg/mouse at ~4 weeks post skin graft, to
A431 tumor-bearing mice 3 weeks post tumor inoculation. Binding intensity was monitored by a Caliper IVIS system.
1500
Heterogeneous
Fig. 1 Altered physicochemical properties of the solid tumor microenvironment.
•
Week
Day
H A L O V a r ia n t 1 - M M A E ( 3 0 m g /k g )
Assessing a Next Generation ADC Technology
Cetuximab
Control
1mm
H A L O V a r a n t 1 - M M A E ( 3 0 m g /k g )
H A L O V a r ia n t 1 - M M A E ( 1 0 m g /k g )
Clinical observations and food consumption
Body weight
Dermal scoring
Clinical pathology
ECG and blood pressure
Veterinary physical examinations and ophthalmology
Histology
Pharmacokinetics
M FI
H+
H A L O V a r ia n t 1 - M M A E ( 1 0 m g /k g )
H A L O V a r ia n t 1 - M M A E ( 3 m g /k g )
6/6 Regressions, no evidence of tumors
0
Day 1
H A L O V a r ia n t 1 - M M A E ( 3 m g /k g )
H A L O V a r ia n t 1 - M M A E ( 1 m g /k g )
2000
M FI
O2
Opportunities for TME Specific Protein Design
↓ pH vs. healthy tissue (acidic pH)
↑ Lactic acid
↑ Albumin
pH 7.4
2B.
Fig. 3 Human skin graft model.
Immunohistochemical analysis
demonstrated EGFR expression in a normal
foreskin xenograft. Human EGFR is visible
in keratinocytes of both the superficial and
basal layer in the xenograft (right arrow).
Mouse skin is negative for human EGFR
(left arrow). No human EGFR staining was
observed in adjacent mouse tissue. The
grafts were stable in mice for >3 months.
40X magnification is shown and 20X
magnification is shown in the inlet.
Staining was performed using a
commercially available anti-human EGFR
antibody.
Proliferation (iododeoxyuridine)
Hypoxia (pimonidazole)
pH 6.5
H A L O V a r ia n t 1 - M M A E ( 1 m g /k g )
2000
M FI
pH 6.0
C e t u x im a b ( 3 0 m g /k g )
•
•
•
•
•
•
•
•
Endpoint
T u m o r V o lu m e ( m m )  S E M
0
V e h ic l e
C e t u x im a b ( 3 0 m g /k g )
3
Blood vessel (immunofluorescent)
10
Cycle 1
2500
V e h ic l e
T u m o r V o lu m e ( m m )  S E M
Solid Tumor
Architechtecture8
20
Parameters
M FI
KRAS mutations present in over 50% of mCRC, predictive of lack of benefit from anti-EGFR4
BRAF mutation in ~10% of mCRC5, debate whether anti-EGFR benefit
EGFR mutation present in about 3% to 19% of NSCLC in Western world, predictive of PFS benefit from anti-EGFR TKI6-7
30
1 Cycle, 2 Dose Study Design, N=3 animals per group (3 each of Vehicle, 2.5 mg/kg, 8 mg/kg groups)
M FI
•
•
•
HALO mAb
Cetuximab
Pilot Toxicology in Non-human Primates
Human CRC Tumor Xenografts
HT29 (BRAFV600E)
2500
(S E M )
• Downstream, Activating Mutations
Skin pH
Acidic pH, TME conditions
40
Fig. 2 Differential in vitro EGFR binding of
Halozyme lead anti-EGFR antibody, HALO mAb.
High affinity was observed for both HALO mAb
and cetuximab at pH 6.0 in high lactate
(16.7 mM) and 25% human serum buffer, a
condition that we approximated as mimicking
the tumor microenvironment in vitro. The
difference in EGFR binding between the 2
antibodies was negligible. However, significantly
attenuated EGFR binding was observed for HALO
mAb at pH 7.4 in low lactate (1.0 mM) and 25%
human serum, when compared to cetuximab.
3
• limits dosing and discourages patients
• ~90% cutaneous side effects1-3
• 8% to 20% severe1-3
50
T u m o r V o lu m e , m m
• EGFR expression leads to serious skin rash
Human TNBC Tumor Xenografts
MDA-MB-231M (KRASG13D)
3
Two Limitations with Anti-EGFR Therapeutics
ADC Targeting of KRAS or BRAF Mutant EGFR+ Tumors
T u m o r V o lu m e , m m
INTRODUCTION
Engineering a TME-Specific Anti-EGFR mAb
EGFR Binding
(EC50 ng/mL)
Cancers with downstream activating KRAS or BRAF mutations in the EGFR pathway are resistant to EGFR targeting agents such as
cetuximab and correspond to a significant unmet need. We hypothesized that an anti-EGFR ADC could be effective against KRAS
or BRAF mutated tumors due to the cytotoxic mechanism of the ADC warhead. In an effort to eliminate the known dermal toxicity
associated with anti-EGFR therapy, and to mitigate potential toxicities associated with treatment by an anti-EGFR ADC, a mAb was
engineered with increased tumor microenvironment (TME) specificity for EGFR. The lead mAb demonstrated undetectable in vivo
binding to human donor foreskins grafted onto nude mice, while binding to human A431 tumor xenografts with similar intensity
to cetuximab (P < 0.005, detected using DyLight-755 conjugated versions of each mAb, measured with a Caliper IVIS system). The
lead mAb was further optimized and conjugated to the potent cytotoxic drug MMAE using a novel bis-alkylating conjugation
linker, which covalently re-bridged the inter-chain disulfide bonds, creating a stable and defined ADC. The resulting ADC, HTI1511, incorporated a vc-PAB cleavable moiety and a short linear PEG (24 ethylene glycol units) in a side-chain configuration.
Analytical HIC revealed that HTI-1511 possessed a nearly homogenous drug:antibody ratio (DAR) of 4 (>99.7%). Approximately
70% of this compound was rapidly internalized by human tumor cells grown in vitro over 4 hours, overlapping the internalization
kinetics of the unconjugated mAb. HTI-1511 was evaluated for efficacy against two human EGFR overexpressing tumor models,
MDA-MB-231M (triple-negative breast cancer, KRAS-G13D) and HT-29 (colorectal cancer, BRAF-V600E), and dosed at 5, 10, and 15
mg/kg, (qw, IV). A clear dose dependent anti-tumor response was observed with complete tumor regressions observed at the 15
mg/kg dose in both models, which were resistant to treatment by cetuximab. In addition, HTI-1511 was well-tolerated at 2 and 8
mg/kg in a cynomolgus monkey toxicity study (n=3 per group), with limited dermal findings that were comparable with the
vehicle control group. No adverse findings were observed at either dose. HTI-1511 showed a high degree of circulating stability in
cynomolgus monkeys, and lacked in vivo degradation and instability that was observed in a control ADC conjugated using
maleimide chemistry. HTI-1511 demonstrated significantly attenuated binding to FcγRIIa, FcγIIb, FcγIIIa 158V, and FcγIIIa 158F
receptors, but not attenuated binding to FcγR1, in a FACS based assay format specific for each receptor, suggesting that HTI-1511
might have improved tolerability due to lack of binding by FcγRII-III receptors, possibly due steric hindrance from the PEG side
chain. Thus, HTI-1511 holds promise as a potentially safe and effective treatment of EGFR overexpressing tumors with KRAS or
BRAF mutations.
40
50
0
20
40
S tu d y D a y
60
80
0
10
20
30
40
50
60
70
80
90
100 110
S tu d y D a y
Fig. 9 Comparison of potency between HTI-1511 and HALO anti-EGFR mAb conjugated with 1st generation chemistry
(maleimide vc-PAB-MMAE). Each ADC was dosed weekly at 5,10, and 15 mg/kg, weekly. HTI-1511 showed superior
potency across all groups. n=6 mice per group. Arrows indicate doses.
1) N Engl J Med. 2004 Jul 22;351(4):337-45. 2) J Clin Oncol. 2012 Oct 1;30(28):3499-506. 3) J Clin Oncol. 2008 Apr
1;26(10):1626-34. 4) Cancer Discov. 2014 Nov;4(11):1269-80. 5) Biomark Cancer. 2015 Sep 6;7(Suppl 1):9-126. 6)
Ann Oncol. 2013 Sep;24(9):2371-6 7) J Natl Cancer Inst. 2013 May 1;105(9):595-605. 8) Lancet Oncology, 2010; 11:
661–69. 9) Thanos CD, Springer, 2016, in press. 10) Bioconjugate Chemistry 2014 Jun 18;25(6):1124-36.