Perspectives on Drug-induced Vascular Injury

Drug-induced vascular injury in nonclinical toxicity studies has been a difficult problem for the pharmaceutical industry for decades. Drugs associated with vascular injury include several classes of antibacterial agents, antimalarials, antivirals, anti-inflammatory agents, and molecules manipulating multiple mechanisms for neurologic diseases, cardiovascular diseases, and diabetes mellitus. While some mechanisms of action (e.g., dopamine receptor agonists or antagonists, phosphodiesterase inhibitors, adrenergic agonists and antagonists, endothelin antagonists, and compounds that modulate endothelial nitric oxide synthase) would be expected to have vasoactive properties, other compounds can produce vascular injury with no obvious mechanism. Acute or multidose cardiovascular safety pharmacology studies in animals often show no systemic effects on blood pressure, heart rate, electrocardiographic data, or cardiac contractility for chemicals associated with vascular injury. Changes in vascular tone in regional arterial beds often are suspected but are difficult to demonstrate. The relevance of these findings in rats, dogs, or monkeys for human risk assessment is uncertain, and drug-induced vascular injury in humans related to vasoactive properties of chemicals is sparsely reported and poorly characterized. Human risk assessment is further complicated because vascular injury is difficult to detect or monitor in patients. Historically, drug-induced vascular injury has been a problem for small molecule pharmaceuticals, but more recently some monoclonal antibodies, other protein therapeutics, and oligonucleotide therapies have produced vascular injury in animal species through mechanisms that are likely different from those for small molecules.

Further complicating risk assessment for vascular injury observed in nonclinical studies is the infrequent and sporadic nature of the findings in animals treated with some chemicals. Spontaneous arterial injury (arteriopathy) in control animals is uncommon, but not rare. The incidences of vascular injury in control rats, dogs, and cynomolgus monkeys from 2008 to 2013 in studies sponsored by Pfizer Inc. are shown in Table 1. Although the incidences vary considerably from year to year, it is unusual to find more than 1 control animal with vascular injury in any particular study, and very often the vascular injury in control animals is minimal in severity and limited to 1 or very few blood vessels. Unfortunately, vascular injury in the treated groups often is similarly limited to 1 or a few animals and only a few blood vessels. Based on the authors’ review of nonclinical toxicity studies over a 5-year period, vascular injury was reported in studies representing 114 small molecule drug candidates across many different mechanisms of action. Fifty-five of these compounds had vascular injury reported in only a single animal in just 1 study, and for 17 of these 55 molecules the vascular injury was limited to the nonrodent species with no vascular findings in rats. When the number of animals and number of blood vessels affected are very low in a study, the determination of whether these effects are test article related or incidental background noise can be difficult, particularly when there are limited historical data for the compound. When the interpretation of small exploratory studies is equivocal, one option is to move forward into larger studies to help provide more definitive information. However, due to the difficulties in differentiating spontaneous from drug-induced vascular injury, some sponsors may halt development of a drug candidate even though it may have benefit without causing vascular injury in patients.

Table 1.

Incidence of vascular injury in control animals in nonclinical toxicology studies up to 3 months in duration (2008–2013) sponsored by Pfizer Inc.

Species	Incidence of vascular lesions in control animals (%)	Range of annual incidences (%)
Rat (Crl:WI and Crl:SD)	1.08	0.26–1.67
N = 5,293
Beagle dog (Marshall Farms)	2.18	0–5.5
N = 597
Cynomolgus monkey (Mauritius origin)	3.43	0–8.11
N = 464

The pharmaceutical industry has been looking for decades for reliable noninvasive biomarkers of vascular injury in body fluids, other methods to detect or visualize vascular injury in humans, tools to identify mechanisms of vascular injury, and in vitro models that accurately predict human vascular injury. Progress has been frustratingly slow. Investigators representing many pharmaceutical companies are collaborating through the Vascular Injury Working Group within the Critical Path Institute’s Predictive Safety Testing Consortium (PSTC) to summarize the mechanisms of arterial injury in animals associated with vasodilation and vasoconstriction, establish a recommended histomorphologic lexicon based on anatomic site of vascular injury, propose a strategy to identify and qualify vascular injury biomarkers, and initiate early testing of multiple proposed biomarkers (see Mikaelian et al. in this issue). The Safer and Faster Evidence-based Translation (SAFE-T) collaboration within the Innovative Medicines Initiative sponsored by the European Commission and the European Federation of Pharmaceutical Industries and Associations (EFPIA) has proposed a similar biomarker strategy in human clinical trials (see Lawton et al. in this issue). These groups are working together and partnering with sponsoring companies to identify and qualify biomarkers that demonstrate and characterize risks for drug-induced vascular injury in animal toxicity studies and patients (see Brott et al. in this issue).

This issue of Toxicologic Pathology reviews much of the current knowledge of drug-induced vascular injury observed in nonclinical toxicity studies in rodents, dogs, and monkeys, as well as strategies to understand and explore this difficult problem. In addition to development and qualification of circulating biomarkers, new strategies are being developed to elucidate mechanisms of vascular injury and better visualize microvascular structures. Mechanisms of vascular injury associated with dipeptidyl peptidase 4 inhibition and the use of mechanistic data to improve human risk assessment are described by Hoffmann et al., while Tobin et al. examine the role of phosphorylation of endothelial nitric oxide synthase in producing arterial injury caused by vasoactive compounds. Arterial injury associated with drug-induced hypoglycemia is reported by Pettersen et al. for the first time. Swanson et al. and Ramot et al. describe recent advances in imaging technology that can improve the nonclinical assessment of vascular injury.

As mentioned previously, drug-induced vascular injury is no longer confined to small molecules. The development of many biologic therapeutics, including monoclonal antibodies, engineered antibody fragments, cytokines, and antibody–drug conjugates containing cytostatic or cytotoxic components, has been terminated or complicated as a direct result of vascular and/or glomerular injury. Mechanisms of vascular injury associated with biologics (when they can be identified) are generally different than mechanisms that directly affect vascular tone. Vascular injury associated with biologics may be caused by direct interaction with a vascular target, antidrug antibody–drug complexes, complement fixation, or leukocyte activation. Diagnosis and mechanisms of tissue injury caused by immune complexes including antidrug antibodies are reviewed in this issue by Rojko et al., while Heyen et al. describe a case study of a monoclonal antibody that produced immune complex disease in individual monkeys resulting in severe clinical signs and death, and the scientific and regulatory strategy that enabled continued development of the compound.

This issue provides readers with a current perspective on much that is known about drug-induced vascular injury in nonclinical toxicity studies and what remains to be accomplished to provide confidence in human risk assessment. Extensive efforts within the pharmaceutical industry are paving the way to maximize identification of real vascular injury risk in patients while decreasing attrition of beneficial drugs that cause vascular injury in nonclinical species, but do not present a significant risk of vascular injury in people. After decades of effort, novel cellular, biochemical, genomic, and imaging biomarkers of vascular injury appear to be on the horizon and offer hope for better human risk assessment in the future.