Re: “Pharmacokinetic Analysis of Intraocular Penetration of Latanoprost Solutions with Different Preservatives in Human Eyes” by Sekine et al. ( J. Ocul. Pharmacol. Ther . 2018;34:280–286)

To the editor:

The population pharmacokinetic report by Sekine et al. ¹ on the comparative aqueous humor bioavailability of several marketed formulations of latanoprost ophthalmic solution is a major opus and provides unique data previously unavailable in the literature. Because of the challenges in conducting clinical ocular pharmacokinetic studies, ² in the U.S., generic ophthalmic solutions (Abbreviated New Drug Application [ANDA]) are approved on the basis of in vitro pharmaceutics alone. ³ However, generic systemic ANDAs are approved on the basis of comparative bioavailability in blood. Typically, the key pharmacokinetic measures for a generic product (maximum concentration, C_max; half life, t½; and area-under-the-curve [AUC]) are required to be within 80%–125% of the innovator product. ^* Most relevant of these pharmacokinetic measures for this once-daily drug is probably the AUC over 24 h (AUC₂₄), rather than for the first few hours after dosing. While the authors provide some of these analyses, they did not seem to calculate the ratio to the innovator, including confidence intervals around this ratio.

The clinical relevance of the present study is challenging to determine. The authors state that “…sample sizes were determined by taking latanoprost free acid into consideration, time to reach the maximum aqueous humor concentration (t_max 5 min) and the elimination half-life (t_1/2, 17 min),” citing a Pfizer document, not generally available to most readers of this journal. I assume that this reflects the selection of time points and enrollment of more patients for the earlier time points to gain more precision. However, they do not define a priori a clinically significant difference (eg, 25% of the C_max as reported in a previous article, ∼17 ng/mL, ⁴ or the 80%–125% range noted above). Thus, we are left with only statistical significance, which is a function of the variability and sample size. Furthermore, the article states “to prevent administration error, patients scheduled to undergo surgery on the same day were assigned to receive the same medication.” I believe this implies that some patients participated in the study with both eyes. This creates both a pharmacokinetic problem (the potential for crossover between eyes) and a statistical problem (eyes covary, and a statistical adjustment needs to be made for eyes within patients). ⁵ It is not possible to evaluate the magnitude of this issue from the data provided.

A major potential for use of this data is to explore the relationship between the pharmacokinetics (how much drug is present) and the pharmacodynamics (what the drug does), known as pharmacokinetic/pharmacodynamic (PK/PD) relationships. No intraocular pressure data are given for these patients, most likely because they underwent cataract surgery within a short interval on the same day. Furthermore, PK/PD relationships for latanoprost are challenging, even with animal AND human data. ⁶ Unlike timolol and pilocarpine, where the time course of ocular hypotensive effect seems to parallel the dosing (ie, peak effect seen within the first few hours after instillation and duration relatively short), prostaglandins have a peak efficacy of 12 or more hours after dosing. ^** Thus, the lag between C_max and peak effect might question whether the C_max is needed for peak efficacy, or rather it is a consequence of needing aqueous levels adequate to sustain a duration of action of 24 h to support a once-daily dosing regimen.

The study might have been more useful with an attempt to measure both parent drug (latanoprost) and active metabolite (latanoprost free acid) in tears, as well as blood.

The authors are to be congratulated on a major work, in which statistically significant differences were seen between formulations in aqueous humor bioavailability of the active metabolite at early postinstillation time points. However, the clinical significance of this finding on ocular hypotensive efficacy is not known.