The collection of aqueous humor (phase 1 b/2 Mahalo study) from patients dosed intravitreally with anti-factor D (AFD; FCFD4514S, lampalizumab), a humanized antibody fragment previously under investigation to treat geographic atrophy (GA) secondary to age-related macular degeneration, presented a unique opportunity to examine AFD properties in clinical samples. We investigated AFD stability and target-binding characteristics to set up strategies for engineering and evaluating optimized molecules that enable less frequent dosing. Two variants, AFD.v8 and AFD.v14, were evaluated as alternatives to AFD for longer-acting treatments. Mass spectrometry, surface plasmon resonance, and immunoassay were used to assess AFD stability and binding activity in aqueous humor samples from Mahalo patients. In vitro stability and binding activity of AFD, AFD.v8, and AFD.v14 were assessed in human vitreous humor versus buffer at 37 °C over 16 weeks and in vivo in rabbits over 28 days along with pharmacokinetic determinations. In human aqueous humor, AFD specific binding was >85% through 30 days, and deamidation was <3% through 60 days, consistent with the AFD stability and binding activity in vitreous humor from humans in vitro and rabbits in vivo. Target binding, stability, and rabbit pharmacokinetic parameters of AFD.v8 and AFD.v14 were similar to those of AFD. Physiological stability and activity of AFD translated across in vitro and in vivo studies in humans and rabbits. The two variants AFD.v8 and AFD.v14 demonstrated comparable potency and pharmacokinetics. These findings, along with previously demonstrated improved solubility of AFD.v8 and AFD.v14, provide proof-of-concept for developing other similar long-acting therapeutic variants.
Keywords: age-related macular degeneration; anti-factor D; aqueous humor; geographic atrophy; human; lampalizumab; long-acting delivery; pharmacokinetics; rabbit; stability; target binding; vitreous humor.