Preclinical Safety Assessment of mAbs
Preclinical safety assessment of mAbs offers unique challenges as many of the classical evaluations employed for small molecules are not appropriate for protein therapeutics in general and mAbs in particular. For example, in vitro genotoxicology tests such as the Ames and chromosome aberration assays are generally not conducted for mAbs given their limited interaction with nuclear material and the lack of appropriate receptor/target expression in these systems. As mAb binding tends to be highly species-specific, suitable animal models are often limited to non-human primates and for this reason, many common in vivo models such as rodent carcinogenesis bioassays and some safety pharmacology bioassays are not viable for mAb therapeutic candidates. For general toxicology studies, cynomolgus and rhesus monkeys are most commonly employed and offer many advantages given their close phylogenetic relationship with humans; however, due to logistics, animal availability, and costs, group sizes tend to be much smaller than typically used for lower species thus limiting statistical power. In some cases, alter-native models are employed to enable studies in rodents. Rather than directly testing the therapeuticcandidate, analogous monoclonal antibodies that can bind to target epitopes in lower species (e.g., mice) can be engineered and used as a surrogate mAb for safety evaluation (Clarke et al., 2004). Often the antibody framework amino acid sequence is modified to reduce antigenicity thus enabling longer-term studies (Albrecht and DeNardo, 2006; Weiner, 2006; Cohenuram and Saif, 2007). Another approach is to use transgenic models that express the human receptor/target of interest (Bugelski et al., 2000); although, results must be interpreted with caution as transgenic models often have altered physiology and typically lack historical background data for the model. To address development issues that are specific to monoclonal antibodies and other protein therapeutics the International Conference of Harmonization (ICH) has developed guidelines spe-cific to the preclinical evaluation of biotechnology-derived pharmaceuticals (ICH, 1997a).
For general safety studies, species selection is an important consideration given the exquisite species-specificity often encountered with mAbs. Model selection needs to be justified based on appropriate expression of the target epitope, appropriate binding affinity with the therapeutic candidate, and appro-priate biologic activity in the test system. To aid in the interpretation of results, tissue cross reactivity studies offer the ability to compare drug localization in both animal and human tissues. For mAb therapeutic candidates, a range of three or more dose levels are typically selected to attain pharmacologically relevant serum concentrations, to approximate levels antici-pated in the clinic, and to provide information at doses higher than anticipated in the clinic. For most indications, it is important to include dose levels that allow identification of a no observable adverse effect level (NOAEL). If feasible, the highest dose should fall within the range where toxicity is anticipated; although, in practice many monoclonal antibodies do not exhibit toxicity and other factors limit the maximum dose. To best reflect human exposures, doses are often normalized and selected to match and exceed anticipated human therapeutic exposure in plasma, serum or blood based upon the exposure parameters, area under the concentration-time curve (AUC), maximum concentrations (Cmax) or concentra-tion prior to next treatment (Ctrough). The route of administration, dosing regimen and dosing duration should be selected to best model the anticipated use in clinical trials (ICH, 1997b).
To adequately interpret non-clinical study results it is important to characterize ATA responses. For human mAbs, ATA responses are particularly promi-nent in lower species but also evident in non-human primates albeit to a lesser degree, making these species more viable for chronic toxicity studies. ATAs canimpact drug activity in a variety of ways. Neutralizing ATAs are those that bind to the therapeutic in a manner that prevents activity, often by inhibiting direct binding to the target epitope. Non-neutralizing antibodies may also indirectly impact drug activity, for example rapid clearance of drug-ATA complexes can effectively reduce serum drug concentrations. In situations where prominent ATA responses are ex-pected, administration of high dose multiples of the anticipated clinical dose may overcome these issues by maintaining sufficient circulating concentrations of active drug. To properly interpret study results it is important to characterize ATA incidence and magni-tude as the occurrence of ATA responses could mask toxicities. Alternatively, robust ATA responses may induce significant signs of toxicity such as infusion-related anaphylaxis that may not be predictive of human outcome where ATA formation is likely to be less of an issue. If ATA formation is clearly impacting circulating drug levels, ATA positive individuals are often removed from consideration when evaluating PK parameters to better reflect the anticipated PK in human populations.