Dexamethasone Dosing of Human Perilymph Compared for Common Delivery Protocols using Inner Ear Simulations

Abstract

Introduction:

Locally applied dexamethasone therapy of the inner ear is commonly used to treat disorders such as Meniere’s disease and idiopathic sudden sensorineural hearing loss. Dexamethasone is also being used in conjunction with cochlear implantation to reduce electrode impedances and fibrous tissue formation around the implant. The variety of formulations currently in use make it difficult to compare the dosing levels resulting from different therapies.

Methods:

We have used the FluidSim cochlear fluids simulation program to calculate perilymph and plasma dexamethasone levels achieved by a variety of dosing approaches used clinically in humans. Identical pharmacokinetic parameters for perilymph and plasma were used for each delivery condition.

Results:

Each of the delivery protocols were calculated to generate therapeutic concentrations in perilymph, but the time course of influence differed markedly between them. Some protocols generate concentrations in the blood that are above therapeutic concentration, increasing the potential for systemic side effects.

Discussion:

Detailed simulations of delivery procedures allow different approaches to be compared quantitatively, giving a measure of dosing efficiency and allowing the merits of each protocol to be compared. Cochlear implants provide the most efficient dosing, generating therapeutic concentrations in the cochlea with minimal systemic influence. For intratympanic delivery, Spiral Therapeutics SPT-2101 provides therapeutic concentrations in perilymph with minimal systemic dosing. Conventional IT therapy with dexamethasone phosphate provides brief therapeutic concentrations in perilymph but substantial systemic exposure, with plasma concentration calculated to exceed therapeutic levels.

Introduction

The current analysis is based on calculated dexamethasone distribution resulting from a number of clinically used protocols in which drug is applied locally to the human inner ear. Based on the amount, rate and location of drug delivered, the time course of dexamethasone concentration achieved in perilymph and blood plasma were calculated. C_max and area under the curve (AUC) over a 28-day period were determined for both perilymph and plasma. Therapeutic concentrations in the systemic circulation resulting from the therapy can potentially cause unwanted side effects. Minor side effects from systemic dosing can include weight gain, mood changes, elevated blood sugar, hypertension and skin changes. More serious side effects from longer exposures include adrenal suppression, myopathy, eye problems, cardiovascular risk, and in rare cases, avascular necrosis. An efficient delivery method would result in a prolonged therapeutic concentration in perilymph while minimizing systemic (blood) concentration. A measure of inner ear dosing efficiency was calculated as the ratio between perilymph and blood C_max values.

The FluidSim program has been used extensively in the design and interpretation of experimental drug delivery studies, with over 40 publications comparing simulations with measured data [1]. The calculations here were all performed with FluidSim v5.0, available for download via the Turner Scientific Website (http://turnerscientific.com/). Perilymph and middle ear pharmacokinetics of dexamethasone and dexamethasone-phosphate have been studied extensively in guinea pigs, with analysis and interpretation based on simulations of the experiments [2–7]. The limited range of dexamethasone measurements available from humans [8,9] have also been interpreted using FluidSim simulations [6,9]. Based on these calculations we have established appropriate pharmacokinetic parameters for dexamethasone, allowing dexamethasone distribution to be replicated with time for a variety of application and perilymph sampling conditions. These studies also support the concept that drug elimination from the middle ear and perilymph is ultimately directed to the vasculature and that systemic drug levels represent a balance between influx from these sources and elimination from the blood [9].

Methods

FluidSim calculations representing the human ear make use of a detailed representation of human inner ear anatomy obtained from 3-D thin-sheet laser imaging microscopy (TSLIM). The cross-sectional area of each fluid and tissue compartment of the human inner ear was defined in 0.1 mm segments along the length of each. The cochlear compartments include scala tympani (ST), spiral ganglion, organ of Corti, the endolymphatic space, spiral ligament, and scala vestibuli including the vestibule (SV). The vestibular compartments include endolymphatic and perilymphatic compartments for each of the semicircular canals, as well as the endolymphatic spaces of the utricle and saccule. Drug distribution along the cochlea is calculated between adjacent 0.1 mm elements of each structure, upon which is superimposed other physical processes such as entry from the middle ear or elution from a cochlear implant.

Pharmacokinetic (PK) parameters for dexamethasone are based on the experimental PK studies in animals outlined above and summarized in Table 1. Unless specified otherwise, these parameters were used consistently for calculations of all delivery conditions. Note that the elimination from the middle ear and from scala tympani and scala vestibuli perilymph (shown bold) is assumed to be ultimately directed to the vasculature, providing the source for the increase in plasma concentration.

Table 1:

Middle and Inner Ear Pharmacokinetic Parameters for Dexamethasone.

Dexamethasone diffusion coefficient	0.77 × 10−9 m2/s
Dexamethasone solubility	89 μg/mL
RWM permeability	153 × 10−9 m/s
Stapes entry (% of total)	50 %, 0% for SPT-2101
Middle ear elimination half time of dissolved drug	28 min
Perilymph, scala tympani elimination half time	40 min
Perilymph, scala vestibuli elimination half time	87 min
Perilymph-endolymph exchange half time	60 min
ST communication half time with adjacent tissues	6 min
Spiral ligament – SV communication half time	120 min
Perilymph-CSF exchange oscillation	3 nL/s
CSF flow into perilymph	30 nL/min

The delivery conditions considered for this analysis include 3 formulations injected intratympanically and dexamethasone eluted from cochlear implants inserted into scala tympani.

i). Intratympanic dexamethasone-phosphate (DexP) solution

This is the most commonly applied formulation delivered locally to the inner ears of human patients. A solution of DexP, as generally formulated for intravenous delivery and used “off-label”, is injected through the tympanic membrane. Typically, the middle ear is filled with solution, requiring about 0.8 mL of drug solution. The concentration applied has varied from 4 mg/mL [8,10] up to 24 mg/mL [11], so we present calculations covering this range. Although highly water-soluble, DexP is a pro-drug and does not bind to glucocorticoid receptors until the phosphate moiety is cleaved through the action of phosphatases.

ii). Intratympanic dexamethasone (Dex) suspension in poloxamer gel

The base form of Dex is far less water soluble than DexP and has been delivered as a suspension of the micronized solid drug which dissolves slowly over time. Extensive studies were performed with intratympanic applications of Dex suspension in poloxamer gel, initially called OTO-104 and subsequently marketed as “Otividex”. [12]. The marketed formulation contained 6% Dex. Approximately 200 uL of the formulation was injected into the round window niche area of humans.

iii). Intratympanic dexamethasone (Dex) suspension in crosslinked gel

This formulation also contains a 6% suspension of Dex but is injected as a crosslinked gel that chemically sets within minutes of injection. Termed SPT-2101, it allows a smaller volume (50 uL) of formulation to be placed directly on the round window membrane. The injected gel remains in place far longer than the prior formulation in poloxamer gel [9]. In previous injection protocols it was assumed that the applied drug is dispersed widely around the middle ear, so we included both entry at the round window and entry in the vicinity of the stapes in simulations [13]. In the case of SPT-2101 the volume is small and is precisely located only on the round window, so entry at the stapes is not included in the simulations.

iv). Dexamethasone elution from cochlear implants inserted into scala tympani.

Cochlear implants are manufactured containing crystalline Dex that elutes slowly over time. However, due to manufacturing differences, there is variation in how much total drug is included in an implant and in the rate at which the drug elutes. In their patent filing, MedEl describe implants with a coating that contains 15.75 μg of dexamethasone that can elute at various rates, depending how the coating is prepared [14]. Over a 28-day period the amount eluted could vary from 35% of the total (5.5 μg) to 65% of the total (10.7 μg) (from their Figure 5). Elution followed a nonlinear curve in which the elution rate declined progressively as a function of time. Cochlear Ltd (Australia) also gave us access to their proprietary data which included implant dimensions, total drug contained in the implant and elution amounts measured over a 28-day period under standardized conditions in a chamber in vitro. Samples for analysis were taken from the elution chamber at 1, 2, 3, 7, 15, 21 and 28 days. From all the available data sets we calculated the highest and lowest elution rates with time. We then set up FluidSim using appropriate implant dimensions, depth of insertion into scala tympani and using the range of elution rates, from which the time courses for scala tympani perilymph and blood were calculated. Curves are presented representing the range of characteristics expected with cochlear implants.

The amount of drug delivered for each of the conditions described above are summarized in Table 2.

Table 2:

Applied Dexamethasone Amounts with different protocols

	Volume	Concentration	Total Dex Delivered
	μL		μg
IT Dex-P	800	4 to 24 mg/mL	3200 to 19200
Otividex	200	6%, 60 μg/μL	12000
Spiral SPT-2101	50	6%, 60 μg/μL	3000
Implant Human			4.1 to 10.7
Implant Mini-Pig			7.4

Measurement of perilymph dexamethasone concentration in humans after implantation with dexamethasone-eluting electrodes is presently not possible. For this reason, we have included calculations for the same types of dexamethasone-eluting implants placed in the mini-pig cochlea. This is because the mini-pig cochlea is of similar size to the human so that human-sized implants (i.e. implants manufactured for human use and with human drug release specifications) can be inserted and their perilymph and plasma pharmacokinetics studied. This provides the potential for perilymph dexamethasone measurements from mini pigs that can be directly compared with the calculations presented here. Calculations made use of cochlear volume and dimension data for the pig [15].

Plasma concentration calculations are based on extensive perilymph and plasma measurements made in guinea pigs, monkeys and humans with intratympanic application of SPT-2101 [9]. The prior analysis compared perilymph and plasma measurements with FluidSim calculations which closely represented the measured data. We used the same kinetic parameters from this prior study to define the systemic concentration time course, as shown in Table 3. The dexamethasone elimination rate for blood of the mini pig is based on the equation from [16] as detailed in [9].

Table 3:

Parameters for Calculating Plasma Concentrations using FluidSim.

	Human	MiniPig
Body Weight (kg)	70	7
Blood Clearance (L/h)	12.1	3.04
Blood Volume Distribution (mL)	74000	7400
Blood Elim T1/2 (min)	255	102

Results

The perilymph and plasma time courses calculated for a 28-day period resulting from 4 dexamethasone delivery protocols are compared in Figure 1. It is notable that as DexP has high aqueous solubility, the applied concentration can be up to 24 μg/mL. This results in a high perilymph C_max but therapeutic concentration is only present for a few hours because concentration in the middle ear rapidly declines with time [6,9]. Losses from the middle ear to the vasculature, to the lymphatics and to the pharynx via the Eustachian tube all potentially contribute to the plasma concentration time course, which is calculated to result in concentrations well above therapeutic concentration for a short time. In Otividex and SPT-2101 formulations, Dex is present as a suspension of the native Dex molecule, which has a maximum aqueous solubility of around 89 μg/mL, the concentration which is driving entry into perilymph. Perilymph levels are therefore initially lower than with DexP, but for Dex are maintained over the 28-day period as the solid drug continues to dissolve over time. The sustained slow release of Dex over time results in lower plasma concentrations than with DexP. For Dex-eluting cochlear implants, the applied drug is in the form of the low-solubility native Dex. Even though the eluted amount is considerably lower than with intratympanic dosing, the perilymph concentration achieved is comparable to intratympanic dosing as drug is being applied directly to the perilymph, avoiding the substantial losses associated with the middle ear. The amount of drug released is so small that resulting plasma levels, from dexamethasone being eliminated to the cochlea vasculature, are calculated to be well below therapeutic concentrations.

Fig. 1.

Calculated scala tympani (ST) perilymph (left side) and plasma (right side) concentrations (top row) and AUC (lower row) for 4 dexamethasone delivery protocols. IT DexP: Intratympanic dexamethasone phosphate at 24 mg/mL (Red solid) or 4 mg/mL (Red dashed). Otividex: Intratympanic dexamethasone suspension (6%, in poloxamer gel, Orange); SPT-2101: Spiral Therapeutics intratympanic dexamethasone suspension (6%, in crosslinked gel, Purple). Implant: Cochlear implant in scala tympani eluting at high rate (Blue solid) or low rate (Blue dashed). The range over which dexamethasone is thought to become therapeutic in the cochlea is shown shaded green.

The calculated kinetic properties for the application conditions are summarized in Figures 2 and 3. Figure 2 compares the C_max values for perilymph and blood with the amount of Dex delivered in each condition. It is notable that the highest perilymph C_max is achieved with IT DexP but also noteworthy that C_max for blood is substantially higher with DexP as the total amount applied is high and the bulk of the applied drug is quickly lost from the middle ear to the vasculature. Blood levels of Dex are likely to well exceed therapeutic concentrations. As a comparison, the total amount of DexP given in the 24 mg/mL protocol amounts to about 19.2 mg which is on the high side of the range of that given IV daily for for inflammation (0.75 – 9 mg), cerebral edema (10 – 24 mg), shock (20 – 40 mg) or allergic reaction (4 – 8 mg) (data from Medscape.com and Drugs.com). Even though it is given intratympanically, the resulting systemic dosing is likely to be significant with high dose IT DexP protocols.

Fig. 2.

Calculated perilymph and plasma C_max for 4 delivery paradigms in humans (detailed in Figure 1) and for a drug-eluting implant in the mini-pig for comparison. The left column shows the total amount of dexamethasone applied in each condition.

Fig. 3.

Calculated perilymph and plasma Area Under the Curve (AUC) after 28 days for 4 delivery paradigms in humans (detailed in Figure 1) and for a drug-eluting implant in the mini-pig for comparison. The left column shows the total amount of dexamethasone applied in each condition.

Dex suspensions (Otividex and SPT-2101) are calculated to generate slightly lower perilymph Cmax values with substantially lower C_max for blood, due to the low solubility of this form of the drug. SPT-2101 is calculated to generate a similar perilymph level as Otividex, but is achieving this with a lower applied amount, reducing the amount that is lost to the blood and resulting in lower blood levels. For SPT-2101 the perilymph and blood levels are supported by measurements in guinea pigs and monkeys and more importantly, the calculated blood levels are well supported by measurements of blood samples from humans with round window SPT-2101 dosing [9].

For Dex delivered from a cochlear implant, the total amount of drug delivered is 20 – 50x lower than with intratympanic dosing. They are calculated to generate similar perilymph concentrations but with far less systemic exposure. With implants, blood levels of Dex are well below likely therapeutic concentrations and are probably below the typical limits of detection with HPLC/Mass Spec. Human cochlear implants placed in minipigs are calculated to generate higher perilymph concentrations, as scala tympani has a narrower cross-sectional area in the pig. They will also generate higher blood concentrations due to the lower body weight, offset to some degree by the faster plasma kinetics for the mini pig (Table 3).

Calculated “Area under the Curve” (AUC) measurements over 28 days are summarized in Figure 3. The amount delivered is again given at the left side for comparison. Intratympanic delivery of DexP results in lowest AUC for perilymph, even with the 24 mg/mL protocol. All of the Dex suspension and elution protocols produce similar overall Dex exposure except for the implants with lower elution rates. AUC for the blood is still highest for 24 mg/mL DexP. The Spiral SPT-2101 generates lowest blood exposure for intratympanic applications. Blood exposure from cochlear implants is approximately 100x lower than with any of the intratympanic applications.

One of the main goals of drug therapy is to deliver drug effectively to a target tissue while minimizing indirect actions and side-effects on other tissues. For local dexamethasone delivery to the inner ear, the ratio of C_max for perilymph and blood were calculated as a quantitative index of dosing efficiency. By this metric, IT DexP is the least efficient, with a ratio of 84.5. Intratympanic Dex suspensions rank as an order of magnitude more efficient with ratios of 418 for Otividex and 1890 for SPT-2101. Drug-eluting cochlear implants are even more efficient with a ratio of 201,000 in minipigs and 641,000 in humans, providing the highest therapeutic perilymph dosing with minimum systemic exposure. A similar analysis was performed using AUC data, but the relationships were similar, so they are not included here.

Discussion

Although Dex and DexP are the most widely used drugs for local treatment of inner ear disorders, there have been no quantitative comparisons of how perilymph dosing differs between the different available therapies. There has only been limited consideration of the unintended systemic dosing with different delivery protocols.

The FluidSim program is well-suited to making such comparisons. The software moves drug between different compartments, including the middle ear, the fluid and tissue compartments of the inner ear and volumes representing the vasculature and cerebrospinal fluid. In all cases drug is only moved around and is never gained or lost in a transfer. The amount subtracted from the source location is added to the destination location. So, once a specific amount of drug is applied, that drug may distribute within the system, but the total amount of drug never changes. The only loss from the system is in the form of elimination from blood (such as mediated by the kidneys) and that amount is documented. This allows us to reliably track how much of the applied drug is in each compartment over time. Measurements of perilymph Dex distribution with intratympanic applications has been analyzed with FluidSim in numerous publications detailed earlier. This understanding of perilymph pharmacokinetics was markedly supplemented by a recent study in which perilymph and blood sample concentrations were compared with FluidSim analysis for SPT-2101 delivery to the ears of guinea pigs and green monkeys. The same formulation was then delivered to humans, followed by the collection of blood samples repeated with time [9]. These data have allowed us to define the relevant kinetic parameters for FluidSim and gives confidence that FluidSim outputs for perilymph and blood concentrations are close to reality.

The most obvious outcome of this analysis is the demonstration of how efficient drug-eluting implanted devices are in achieving therapeutic perilymph concentrations with negligible systemic exposure in humans. The total Dex amount eluted from a cochlear implant is < 1/100th of the amount typically given with intratympanic delivery, yet it achieves similar perilymph concentrations. It demonstrates the unavoidable problem that when drugs are delivered intratympanically, the vast proportion is being lost to the vasculature with only a small proportion entering perilymph. In the prior study [9] (their supplemental material, Figure S3) it was calculated for guinea pigs when 600 μg of dexamethasone was applied to the middle ear in gel formulation, 47% of the total amount dissolved over the 28 day period. Of this dissolved amount, 96% was lost to the vasculature, with just 3.4% entering the inner ear, an amount sufficient to account for the perilymph concentrations achieved. This serves to demonstrate a major problem with intratympanic delivery, specifically that it is a relatively inefficient process, with far more drug lost to the vasculature than that entering perilymph.

Nevertheless, SPT-2101 is shown to be the most efficient intratympanic formulation. Unfortunately, the most common practice of delivering DexP to the middle ear as a solution appears to be the least efficient delivery option. It does provide perilymph dosing for a brief period but also results in substantial systemic exposure to the drug.

Based on prior studies [6,9], we think it is justified to assume that drug loss from the middle ear becomes distributed systemically, driving the amount present in the blood. The blood data for guinea pigs, monkeys and humans with SPT-2102 applications are consistent with this assumption. However, it is accepted that there are multiple pathways for drug loss from the middle ear. With the SPT-2102 formulation, loss may occur directly to the vasculature of the middle ear mucosa, but may also occur via the lymphatic system [17], ultimately entering the vasculature at one of the cervical lymph nodes. With the SPT-2101 formulation we would expect little or no loss via the Eustachian tube to the pharynx as the gel matrix cannot be swallowed. In contrast, DexP administered as a solution may be more likely to be lost to the pharynx and swallowed once the patient resumes normal upright posture after injection. DexP is polar and based on its properties will not be absorbed easily by the gut. But within the gut there is an array of phosphatases, such as intestinal alkaline phosphatase, that will cleave the phosphate group from the molecule. The remaining free dexamethasone is small, lipophilic and readily taken up by the gut. Oral biovailability has been calculated to be approximately 81% [18], accounting for why free Dex is used in oral formulations rather than DexP. Nevertheless, in the absence of blood measurements in patients with IT DexP injections it remains possible that FluidSim may be overestimating the blood levels for this condition to a degree.

The analysis provided by FluidSim may become increasingly useful as new formulations and delivery systems are developed for local delivery to the inner ear. It allows the specific characteristics of the delivery system to be “plugged in” and compared with prior protocols before expensive preclinical studies and clinical trials are considered.

It is accepted that there are always going to be limitations to computer simulations of a biological system. Most of the kinetic parameters used by FluidSim have been established by PK studies in animals. Where available, data from humans [8] have been compared and the results compare favorably with FluidSim simulations [6] (their figure 8a). As we accumulate data with large animal models, such as the mini pig, we will be better able to characterize distribution in larger cochleas more comparable to the human. Nevertheless, quantitative analyses, such as these, provide an important theoretical backbone which can be refined with time as more data are accumulated.

Data Availability Statement

All data generated during this study are shown in the article. The raw data, including setup files for each FluidSim configuration are available on request and will be provided by the authors without reservation. Requests should be directed to the corresponding author.