Challenges and opportunities for the future of monoclonal antibody development: Improving safety assessment and reducing animal use

Figures & data

Figure 1. Use of animals in mAb development and change in practice over time.

Table 1. Questions for case studies addressed during the breakout sessions.

1. Were the effects observed in the preclinical studies conducted to support the first-in-human (FIH) study predictable based on the mechanism of action (MOA)?

a) Which effects were predictable based on the MOA? Please describe.

b) Which effects were not predictable? Please describe.

c) Could the non-predictable effects be attributed to either a) a background lesion in the animal used or b) a consequence of immunogenicity in the animal?

d) Do you think the use of a wider range of in vitro approaches may have aided the prediction of the effects observed (both those based on MOA and other non-MOA effects)?

i. If your response is yes, please describe in further detail.

ii. If your response is no, please describe the knowledge and data gaps, or other issues, which make such an in vitro approach problematic.

2. Prediction of clinical dose/exposure: Could human serum levels/PK be predicted using in vitro data, in vivo data from rodents, or a combination of these?

If your response is no:

i. Please describe the knowledge and data gaps, or other issues, which make such an approach problematic.

ii. What are your reasons for requiring data from NHPs to predict the clinical dose/exposure?

3. Could this product have safely entered into clinical trials on the basis of in vitro approaches driven by the MOA?

4. What additional effects were observed in longer-term general toxicity studies that were not observed in the studies conducted to support the FIH study?

a) Which of the additional effects were predictable based on the MOA? Please describe.

b) Which effects were not predictable? Please describe.

c) Did the additional effects impact on clinical decision making during clinical development e.g., design of clinical studies (inclusion, exclusion criteria, clinical doses, clinical monitoring)?

5. Could this product have progressed through clinical development and to registration on the basis of in vitro approaches?

a) If your response is no, could the use of in vitro approaches lead to a reduced need for in vivo studies? Please summarize your discussions.

6. If studies to assess toxicity to reproduction were conducted for the product, were the observed effects predictable based on the MOA?

a) Which effects were predictable based on the MOA?

b) Which effects were not predictable?

c) Could in vitro approaches have provided sufficient information for clinical risk communication and management?

7. In your view are juvenile toxicity studies warranted for this product to support pediatric indications?

8. Could a biosimilar product for this case study be developed fully in vitro?

Table 2. In vivo studies for anti-ADAMTS-5.

Species	Study^1	Dose/s	Group sex and size	Findings
Cynomolgus monkey	4 week DRF	0 mg/kg (iv and sc)	3F/group	Minimal focal endocardial hemorrhage in the left ventricle in 2 of 3 monkeys in 300 mg/kg/dose
		30 mg/kg (sc)300 mg/kg (iv)		Uncertain relationship to test article due to high background incidence
Cynomolgus monkey	8 week GLP toxicology	0 mg/kg (iv and sc)	3M+3F/group	No cardiac findings (including ECG and cardiac troponin)
		30 mg/kg (sc)		Abnormal fecal consistency and decreased inorganic phosphorus concentrations noted in monkeys given 300 mg/kg/dose
		100mg/kg (iv)
		300 mg/kg (iv)
Wistar Han rat	4 week DRF	0 mg/kg (iv and sc)	4M/group 3M/group for TK	No noteworthy findings
		10 mg/kg (iv)
		10 mg/kg (sc)
		300 mg/kg (iv)
Wistar Han rat	8 week GLP toxicology	0 mg/kg (iv and sc)	10M+10F/group 3M+3F/group for TK	Decreased mean body weight gain and food consumption in males given ≥100 mg/kg/week iv.
		30 mg/kg (sc)
		100mg/kg (iv)
		300 mg/kg (iv)
Cynomolgus monkey	Single Dose GLP Cardiovascular/Respiratory	0 mg/kg (iv and sc)	3F/group	Dose dependent ST segment elevation at 30 and 300 mg/kg on Days 1, 7, 14 and 21
	Jacketed ECG and telemetry device for arterial blood pressure	30 mg/kg (sc) 300 mg/kg (iv)		Other ECG waveform abnormalities
				– 30 mg/kg – intraventricular conduction delay and increased frequency of isolated premature ventricular contractions (PVCs, including bigeminy and couplets), occurring mainly on Day 1
				– 300 mg/kg - increased frequency of isolated PVCs and 1 occurrence of R on T PVC on Days 1 and 7
				↑ in mean arterial pressure (MAP)
				– 30 mg/kg - increase on Day 21 with max. increase of ∼31 mmHg or 30% vs vehicle
				– 300 mg/kg - increase on Days 14 and 21 with max. increases of ∼23 mmHg or 22% vs vehicle. This dose also produced increased Cardiac Work with increases of 39–48%
				At ∼7 months post-dose (non-GLP), ST segment elevations still present
				– In heart tissue, there were no test article-related microscopic findings or differences in IHC for Connexin 43 (gap junctions), versican proteolytic fragments (ECM) or GSK2394002
				– MAP was not evaluated because telemetry implants were no longer operational
				No consistent effects on HR or PR, QRS or QTc interval durations
				No effects on ventilatory parameters
Wistar Han rat	Investigative Repeat Dose Cardiovascular	0 mg/kg (iv) 300 mg/kg (iv)	4M/group	Single dose at 300 mg/kg in conscious rat did not produce any waveform abnormalities or arrhythmias (including ST segment changes) or arterial pressure changes on Days 1, 8 or 15 (continuous 24 hr monitoring)
	Telemetry device for arterial pressure, heart rate, ECG (intervals and waveform abnormalities/arrhythmias), and body temperature
Cynomolgus monkey	Repeat Dose GLP Cardiovascular study to identify NOEL	0 mg/kg (iv)	3M+3F/group	0.3 mg/kg
		0.3 mg/kg (iv)		– Increased MAP following the 3rd dose (up to 8 mmHg)
		3 mg/kg (iv)		– One monkey had multiple episodes (31) of non-sustained ventricular tachycardia (NSVT) following the 1st dose
				Three mg/kg
				– Increased MAP following 2nd dose (up to 9 mmHg or 9% of vehicle) and throughout the remainder of the study (Day 70)
				– Decreased HR between 4–16 hrs following 1st and 2nd dose (up to 19 bpm or ∼13%)
				No evidence of ST segment elevation at 0.3 or 3 mg/kg
Cynomolgus Monkey	Vehicle Investigative Study	Veh (iv) or Veh (iv) with 10X Tween™ 80 (0.2% w/v) or 10X L-arginine (10% w/v)	4M+4F (cross over)	No arrhythmic effects with vehicle at 10X Tween™ 80 (0.2% w/v) or 10X L-arginine (10% w/v)

All studies were conducted in accordance with the GSK Policy on the Care, Welfare and Treatment of Laboratory Animals and were reviewed the Institutional Animal Care and Use Committee either at GSK or by the ethical review process at the institution where the work was performed.

Table 3. In vitro studies for anti-ADAMTS-5.

Species	Study	Findings
Rabbit	Rabbit cardiac wedge assay	No arrhythmogenic activity or ST segment elevation observed at concentrations up to 500 µg/mL
Human	Human CV ion channel assays (hERG, NaV1.5 and CaV1.2)	No difference from vehicle at concentrations up to 500 µg/mL (note – acetate inhibited NaV1.5)
	Selectivity profiling	Fully selective for ADAMTS-5 within class (ADAMTS family and MMPs)
		ProtoArray™ screen (>9400 proteins) – Only specific binding to ADAMTS5 observed
		Tissue cross reactivity – no signal, anti-ADAMTS-5 not a good immunohistochemistry reagent

Figure 2. Dose-dependent ST segment elevation with anti-ADAMTS-5 in cynomolgus monkey.

Table 4. In vivo studies for anti-DLL4 IgG1 mAb.

Species	Study	Dose/s	Group sex and size	Findings
Sprague Dawley rat	8 week GLP toxicology	0 mg/kg (iv)	10M+10F/group and 5M+5F recovery/group and 6M+6F TK satellites/group	Liver findings: Hepatic centrilobular to bridging sinusoidal dilation
		1 mg/kg (iv)		Decreased red blood cells (monitorable)
		3 mg/kg (iv)		Proliferative vascular neoplasms
		10 mg/kg (iv)		– Skin, lung, and heart lesions seen only in rats
		30 mg/kg (iv)
Cynomolgus monkey	8 week GLP toxicology with 12 week recovery	0 mg/kg (iv)	3M+3F/group and 2M+2F recovery/group and 2M+2F for telemetry on the control and high dose group	Liver findings: Hepatic centrilobular to bridging sinusoidal dilation
		0.2 mg/kg (iv)		Decreased red blood cells (monitorable)
		0.8 mg/kg (iv)
		3 mg/kg (iv)
		12 mg/kg (iv)

Table 5. In vivo studies for anti-DLL4 Fab’2 fragment.

Species	Study	Dose/s	Group sex and size	Findings
Sprague Dawley rat	8 week GLP toxicology	0 mg/kg (iv)	10M+10F/group and 5M+5F recovery/group and 6M+6F TK satellites/group	Liver findings: Decreased severity compared with IgG, minimal changes observed at low doses (up to 10 mg/kg)
		3 mg/kg (iv)		Acceptable hematology profile
		10 mg/kg (iv)		Proliferative vascular neoplasms not observed
		100 mg/kg (iv)		New findings: Vascular proliferative/degenerative changes in the heart and lung potentially related to pulmonary hypertension
		30 mg/kg (iv)
Cynomolgus monkey	8 week GLP toxicology with 12 week recovery	0 mg/kg (iv)	3M+3F/group and 2M+2F recovery/group and 2M+2F for telemetry on the control and high dose group	Liver findings: Decreased severity compared with IgG
		5 mg/kg (iv)		Acceptable hematology profile
		15 mg/kg (iv)		New findings: Vascular proliferative/degenerative changes in the heart and lung potentially related to pulmonary hypertension
		50 mg/kg (iv)

Table 6. In vivo studies for anti-amyloid β mAb.

Species	Study	Dose/s	Group sex and size	Findings
APP-transgenic mouse model	Histopathological evaluation of select tissues in a pharmacology study.	0.5 mg/mouse/ week (ip)	15M/group; 7-11 animals/group available for histopathological evaluation	Histological changes (neoplasia, renal changes, degenerative joint disease, skeletal muscle de/re-generation) were considered due to aging or potential background changes in mouse strain and not due to antibody treatment.
APP-transgenic mouse model	4 week GLP toxicology	0 mg/kg (iv and sc)	12M+12F/group and 6M+6F recovery/group and 18M+18F TK satellites/group. Plus additional TK groups for control and 200 mg/kg (iv) with 12 weeks treatment.	Up to 200 mg/kg for 4 weeks was well tolerated. 200 mg/kg for 12 weeks was well tolerated in animals for TK analysis. Anti-drug antibodies were detected in 62% of antibody-dosed animals.
		60 mg/kg (iv)		NOAEL = 200 mg/kg.
		200 mg/kg (sc)
		200 mg/kg (iv)
Cynomolgus monkey	PK study	5 mg/kg single dose (iv and sc)		t ½ ∼8 d. Low clearance and volume of distribution. No adverse or notable effects observed.
Cynomolgus monkey	13 week GLP toxicology	0 mg/kg (iv and sc)	5M+5F	Low doses (20 and 60 mg/kg): Thrombocytopenia (decreased platelets), pulmonary vasculature changes (medial hypertrophy, thrombosis), lung findings (multifocal hemorrhages, interstitial fibrosis).
		20 mg/kg (iv)	120 mg/kg (iv) 1M	High doses (120 and 200 mg/kg): Acute infusion reaction (5-10 min post-dose) with lethal consequence at 200 mg/kg.
		60 mg/kg (iv and sc)	200 mg/kg (iv) 4M
		120 mg/kg (iv)
		200 mg/kg (iv)
Cynomolgus monkey	Exploratory toxicity study	0 mg/kg (iv bolus)	1M+1F	IV infusion (1 hour) produced a less severe acute infusion reaction compared with iv bolus administration at the same dose level. Acute infusion reaction ≥ 60 mg/kg included complement activation. No effects on cytokines, coagulation factors or ECG's. Histopathologic effects observable after 4 weeks of dosing (thrombotic and/or arterial changes in brain, lung and injection sites).
		2 mg/kg (iv bolus)		NOAEL = 2 mg/kg
		60 mg/kg (iv bolus)
		120 mg/kg (iv infusion or bolus)
		200 mg/kg (iv infusion)

Table 7. In vitro studies for anti-amyloid β mAb.

Species	Study	Findings
Human	Cytokine release	Cytokine release assay in whole human blood (IL-1ra, IL-1b, IL-6, hTNFa, IL-8) was positive for IL-8 release at ≥ 10 µg/mL.
Human	PBMC binding	No PBMC staining at 10 or 30 µg/mL.
Human, cynomolgus monkey, Tg mice (3 donors each)	Tissue cross-reactivity with FITC-labeled Ab (GLP)	No noticeable binding.
Human and cynomolgus monkey	Platelet binding (FACS)	No binding to platelets in human blood (Fc mutation and wt alike). No binding to CD41+ platelets in cynomolgus blood (Fc mutation and wt alike). Some binding to CD41- “platelet-like” cells in cynomolgus blood (Fc mutation and wt alike).
Human, cynomolgus monkey, mouse and rat	Serum binding (ELISA)	Concentration-dependent binding to serum component of species tested (Fc mutation and wt alike). No binding of murine Ab to any species. Binding to serum component is not CDR/target mediated; suggests that binding is not Fc mediated.

Table 8. Summary of case-study data and the ability of existing in vitro and in silico technologies to detect or predict toxicities observed in the in vivo studies.

	Anti-ADAMTS-5 mAb	Anti-DLL4 mAb	Anti-amyloid β mAb
Were findings based on the mechanism of action of the mAb?	Yes	Yes	No
Were findings predictable?	Although cardiovascular binding was detected the toxicological effects were not predictable	Some findings were predictable but additional clinically relevant findings were detected that were not predictable	No
Is an in vitro model available now or in the near future?	Not for hemodynamics or secondary effects (e.g., extracellular matrix interactions)	Generally no, due to the need to model paracrine effects between multiple cell types	After off-target binding was identified, in vitro screening could be used to select a better molecule without off-target binding, but still need some in vivo studies to test that the lack of off-target binding does not cause in vivo effects.
What was the impact of the in vivo studies?	Clinically relevant toxicity was only identified in the stand alone safety pharmacology study	Data from in vivo studies were used to make a decision to terminate the mAb and Fab’2 programs.	Data from the NHP study identified cross-reactivity that would not have been identified in mouse, rat or dog. Clinical studies in humans were not conducted with this antibody.

Mullard A. 2015 FDA drug approvals. Nat Rev Drug Discov 2016; 15(2):73–6; PMID:26837582; https://doi.org/10.1038/nrd.2016.15

 PubMed Web of Science ®Google Scholar