The study of intravitreal drug pharmacokinetics: does it matter? and if so, how?

KEYWORDS:

• Binding affinity
• half-life
• intravitreal
• pharmacokinetics
• vascular endothelial growth factor

1. Introduction

The development and widespread adoption of drugs that inhibit the actions of vascular endothelial growth factor (VEGF) have transformed the treatment of chorioretinal vascular conditions during the past 12 years. Vitreous injections place the drug close to the retinal pathology, delay its elimination, and limits systemic exposure. Despite the target specificity of depot injections, much of the drug passes from the eye, unaltered, into the systemic circulation.

The drug’s tissue penetration, binding affinity to the target ligand or receptor, and intraocular half-life are thought to be important determinants of peak efficacy and duration of action. For example, rapid binding of the drug to the target molecule is favorable, whereas rapid binding to interstitial proteins ‘consumes’ the drug by preventing its binding to target molecules [1]. Pharmacokinetic studies that are performed early during development identify those drugs with favorable clinical potential and weed out others that are likely to perform poorly. Initial pharmacokinetic studies performed by the developer acquire important information that guides dosing frequency, optimizes clinical efficacy, and supports applications for regulatory approval.

A second wave of pharmacokinetic studies, performed by independent investigators, usually follows initial drug development and clinical testing. Both in vitro and in vivo (performed in animals and humans) studies further characterize the drug’s actions. Clinicians often use data from these studies to predict clinical responses in situations that have not been studied in the pivotal prospective trials.

Since intravitreal injections are often uncomfortable and expensive, and can cause vision-threatening complications, significant interest in predicting efficacy and duration of action has developed. Scores of published papers discuss binding affinities for VEGF, tissue penetration, intraocular half-lives, and extra-ocular distributions. But despite the large volume of accumulated experimental data, substantial gaps in our knowledge regarding clinically applicable pharmacokinetics remain.

2. Nonhuman vitreous studies

Experimental data suggests that removing the matrix of vitreous proteins and hyaluronate increases diffusion velocity of macromolecules up to six-fold [2]. This supports the contention that intraocular drug half-life in vitrectomized eyes is shorted and it challenges clinicians when they are faced with the prospect of performing a vitrectomy that might decrease treatment efficacy or shorten duration of action. This edition of Expert Opinion on Drug Metabolism and Toxicology features a review article on the intravitreal half-lives of aflibercept (Eylea®, Regeneron, Tarrytown, NY, USA), bevacizumab (Avastin®, Genentech, S. San Francisco, CA, USA/Novartis, Basel, Switzerland), pegaptanib (Pfizer, New York, NY, USA), and ranibizumab (Lucentis®, Genentech, S. San Francisco, CA USA/Novartis, Basel, Switzerland) in animal and human eyes that had previously undergone pars plana vitrectomies [3]. The authors conclude that most studies show decreased intravitreal drug half-lives in vitrectomized eyes but results vary with the models used, thereby emphasizing the difficulty in generalizing these conclusions to patient care.

Surgical studies are frequently performed in rabbits because they are relatively inexpensive to purchase and maintain, and can be handled easily. But rabbit crystalline lenses are large relative to the volume of the vitreous cavity so a vitrectomy must be combined with a lensectomy and capsule removal. Since an intact lens/iris diaphragm impedes diffusion of macromolecules from the vitreous into the anterior chamber, performing a simultaneous lensectomy/vitrectomy in rabbits creates an anatomic situation that is not seen in most human eyes.

3. Human vitreous studies

We have limited data regarding the efficacy of intravitreal therapy in human eyes that have undergone vitrectomy, with the best studies emerging from diabetic macular edema (DME) trials. A post hoc analysis of vitrectomized eyes in the Diabetic Retinopathy Clinical Research Network Protocol I trial showed that mean improvements in visual acuity were the same as in non-vitrectomized eyes but more injections were required [4]. The CHAMPLAIN trial showed that the dexamethasone insert (Ozurdex®, Allergan, Irvine, CA, USA), which elutes drug according to a zero-order pharmacokinetic model, produced improvements in macular edema and visual acuity comparable to those from similar trials that did not include vitrectomized eyes [5].

Unfortunately, comparable data is not available for eyes with neovascular age-related macular degeneration (nAMD) and generalization of the DME results may not be possible because the pathophysiologies of the two conditions differ significantly. Considerable evidence supports that notion that pars plana vitrectomy improves DME, probably by increasing oxygen concentrations within the vitreous [6] and removing edema-promoting molecules sequestered in the posterior hyaloid [7]. Therefore, physicians may have a low threshold for performing a vitrectomy in eyes with DME, but they should carefully consider a vitrectomy in eyes with nAMD because pharmacotherapy may be rendered ineffective.

4. Pharmacokinetic data

A considerable amount of data has been accumulated regarding the pharmacokinetic characteristics of anti-VEGF drugs. Reported VEGF₁₆₅ binding affinities among the drugs differ by up to 100-fold (aflibercept > ranibizumab > bevacizumab) [8] and considerable variance is seen for bevacizumab (58–1100 pM) [8,9] and ranibizumab (46–172 pM) [8,10]. Further complicating matters are the facts that measured binding affinities vary with the assays used by investigators and in vitro binding affinities may not translate directly to in vivo biological effects. Predicted and measured intravitreal half-lives differ by up to two-fold [11]. By using these data in mathematical models, durations of action have been predicted to differ between the drugs by up to 2.5-fold [12].

But the reality, based on scores of clinical trials with the three drugs, is that peak efficacies are nearly identical and durations of action differ by less than 1 week. So why is there such a large discrepancy between predictive modeling and clinical observations? The answer to this question is not known.

The large amount of injected drug overwhelms the small number of diffusible VEGF molecules resulting in rapid and complete binding of all VEGF. This shuts down all VEGF actions, which continues until drug concentrations fall to nanomolar levels (after 4–8 weeks). At this time, VEGF actions quickly rise from 0 to 100% with only small additional decreases in drug concentrations [13]. Direct measurements of VEGF inhibition within target tissue is not possible in vivo, but changes in unbound VEGF concentrations within the aqueous humor serve as a reasonable surrogate. Quadrupling the dose of an injected drug does not increase peak efficacy since this only increases the already excessive binding capacity, but it may slightly prolong the duration of action [14].

5. Investigational anti-vegf drugs

Newly developed anti-VEGF drugs – abicipar pegol (Allergan, Irvine, CA, USA) and brolucizumab (Alcon, Ft. Worth, TX, USA) – are in phase III trials for the treatment of nAMD. Based on the apparent long intraocular half-life of the pegylated abicipar (13.4 days in a phase I/II DME trial) and the large injected quantity (6 mg) of the small (molecular weight: 26 kD) brolucizumab molecule, developers of each drug believe that dosing intervals of 3 months may be achievable.

6. Expert opinion

Three-month dosing intervals would be an attractive option for physicians and patients but it’s not clear how physicians would translate phase III trial results into actual practice. Developers have given the experimental drugs an artificial advantage – dedicated 3-month treatment arms – that may incorrectly be interpreted as showing longer durations of action over the control drugs. Critics will justifiably claim that the rigid trial designs prevented aflibercept and ranibizumab (serving as controls) from being dosed less frequently than bimonthly or monthly. This occurred in the VIEW trials when 2-month dosing of aflibercept was as effective as monthly ranibizumab during year 1, but when patients were treated pro re nata (PRN) with a 12-week cap in year 2, the mean difference in durations of action was less than 1 week. We will probably need post-approval data from PRN and treat and extend studies to determine the true durations of action of abicipar and brolucizumab, and to learn if these differ from ranibizumab and aflibercept.

Despite the plethora of post-approval information regarding the pharmacokinetic properties of anti-VEGF drugs, these differences appear to account for very little in terms of peak efficacy and duration of action. The trend among the three available drugs is that the slightly longer duration of action correlates most closely with VEGF binding affinity. If this trend persists, we can expect abicipar (K_D = 2 pM) and brolicizumab (K_D = 1.6 pM) to most closely resemble aflibercept (K_D = 0.45 pM). We welcome the future approval of these two new drugs but I suspect that they will provide little advantage over the currently available drugs.

Declaration of Interest

M Stewart discloses acting as consultant for Alkahest and Bayer and receiving Institutional Research Support from Allergan and Regeneron. The author has no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed. Peer reviewers on this manuscript have no relevant financial or other relationships to disclose.

Additional information

Funding

This paper is not funded