Pharmacokinetics and Pharmacodynamics of LY2599666,a PEG-Linked Antigen Binding Fragment that Targets Soluble Monomer Amyloid-β

Abstract

LY2599666 is a humanized, affinity-optimized monoclonal antibody antigen-binding fragment linked to a PEG molecule and targets soluble amyloid-β (Aβ) monomers. This first-in-human dose ascending study assessed pharmacokinetics (PK) (measured as serum free LY2599666 concentration) and pharmacodynamic (PD) effects (measured as plasma total soluble Aβ₄₀ and Aβ₄₂) after a single subcutaneous (SC) dose of 10, 25, 100, and 200 mg LY2599666 in healthy subjects. As LY2599666 binds to multiple soluble Aβ monomers, a two-target mediated drug disposition model (TMDD) was developed to simultaneously fit serum LY2599666 concentration and Aβ monomer levels. Four Alzheimer’s disease patients completed 25 mg once-weekly dosing of LY2599666 for 12 weeks. In addition, single cerebrospinal fluid samples were collected to assess penetration capability across the blood-brain barrier. PK and PD data collected from the multiple dose cohort aligned with model predictions, suggesting the established TMDD model predicted suppression of soluble Aβ₄₀ and Aβ₄₂ in plasma after SC dosing of LY2599666.

Keywords

Aβ monomer Alzheimer’s disease pharmacokinetic/pharmacodynamic target-mediated drug disposition

INTRODUCTION

Alzheimer’s disease (AD) is a neurodegenerative disorder characterized by a progressive decline in cognitive function with age. Evidence from preclinical, clinical, and genetic studies suggest that the production and deposition of Aβ is an early and necessary event in the pathogenesis of AD [1 –3]. Therefore, treatments that halt the synthesis and deposition of Aβ, or increase its clearance, might have therapeutic benefits for the patients and slow the progression of AD. LY2599666 is a humanized, affinity optimized monoclonal antibody (mAb) antigen-binding fragment (Fab) linked to a polyethylene glycol molecule (PEG), and is expected to bind the mid-domain epitope of the soluble Aβ peptide. Preclinical studies suggest that LY2599666 mediated reduction of soluble Aβ and may slow the progression of AD. This first-in-human, multicenter, Phase I, double-blind, randomized, placebo-controlled study (NCT02614131) was designed to explore safety, tolerability, pharmacokinetics (PK) and pharmacodynamics (PD) effects of a single subcutaneous (SC) dose of LY2599666 (10, 25, 100, 200 mg) in healthy subjects (Part A), and PKPD of multiple doses of LY2599666 in AD patients (Part B). The single ascending dose part (Part A) was completed while the multiple ascending dose part (Part B) was stopped based on the lack of efficacy results of another compound (solanezumab) directed against the same target. Nonlinear PK was observed based on single dose PK data, suggesting that LY2599666 may undergo target mediated drug disposition. Accordingly, a two-target mediated drug disposition model (TMDD) was developed to simultaneously fit PK (free serum LY2599666 concentrations) and PD (total soluble Aβ₄₀ and Aβ₄₂ monomer levels) data from Part A. Four AD patients completed 25 mg SC once-weekly (QW) dosing before Part B was stopped, and the PK PD data were used as external validation for TMDD model. Cerebrospinal fluid (CSF) samples were also collected at a single time point post 12 weeks of 25 mg QW SC administration of LY2599666 to assess its penetration capability across the blood-brain barrier (BBB).

METHODS

Study design

This Phase I study was conducted in two parts. Part A included four cohorts (Cohorts 1 to 4, dose: 10, 25, 100, 200 mg), with 11 subjects in Cohorts 1 and 2 (8 LY2599666:3 placebo) and 10 subjects in Cohorts 3 and 4 (8 LY2599666:2 placebo). After a single SC administration, serum samples for PK and PD analysis were collected for each patient at 0, 2, 4, 8, 12, 24, 48, 72, 96, 120, 168, 216, 264, 336, 504, and 672 h. For Part B, it was intended that 12 patients would be enrolled in the 25 mg SC QW Cohort. However, following the enrollment of 7 patients, a decision was made to stop the development of LY2599666. PK and PD samples in 25 mg QW multiple dose cohorts were collected for each patient at: 0, 2, 4, 8, 12, 24, 48, 72, 96, and 120 h after the first dose, and 0, 2, 4, 8, 12, 24, 48, 72, 96, 120, 168, 216, 264, 336, 504, and 672 h after the last dose on day 85 (13th dose). Lumbar punctures was performed at baseline and before the last dose (Day 84) of study treatment to access trough concentration.

Bioanalytical methods

Human serum samples obtained during this study were analyzed for LY2599666 using a validated ELISA method (Pharmaceutical Product Development, Richmond, VA, USA). The lower limit of quantification was 125 ng/mL, and the upper limit of quantification was 3000 ng/mL.

Human CSF samples obtained during this study were analyzed for LY2599666 using a validated high-performance liquid chromatography method using tandem mass spectrometry detection (Algorithme Pharma, Laval, Québec, Canada). The lower limit of quantification was 0.500 ng/mL and the upper limit of quantification was 100.000 ng/mL.

Aβ peptides in human plasma were first extracted using a solid phase extraction (SPE) method, and were subsequently measured using a validated enzyme-linked immunosorbent assay (ELISA) method. During the SPE procedure, 300μL of plasma with or without LY2599666 was mixed with 700μL of 6.4 M guanidine-HCl to denature proteins and to separate Aβ peptides from LY2599666. The samples were incubated at room temperature with shaking for 1 h, and then were acidified using H₃PO₄ to a final concentration of 1.2% before being applied to an Oasis HLB (60 mg sorbent) 96-well plate (Waters, Milford, MA) with the aid of a Waters Positive Pressure-96 Processor (Waters). During the SPE process, the Low Flow Regulator on the processor was set to 3 psi to ensure slow passage of samples through the plate to maximize Aβ peptide binding. Before the denatured and acidified samples were loaded to the plate, the plate was first conditioned using 1 mL of methanol, and then was equilibrated with 1 mL of 5% methanol with 4% H₃PO₄. After sample loading, the plate was washed twice using 1 mL of 5% methanol with 4% H₃PO₄ and then with 1 mL of 10% acetonitrile. The bound peptides were eluted twice using 400μL of 75% acetonitrile/3% ammonium hydroxide solution. The elute was dried at room temperature using a SpeedVac system for at least 8 h. To minimize nonspecific binding loss of Aβ peptides to the 96-well plate, sucrose was added to a final concentration of 1% before the samples were dried. Control samples with Aβ₄₀ and Aβ₄₂ spiked into plasma pooled from healthy donors, at concentrations of 0.5, 1, and 5 ng/mL, were also processed in parallel. Before ELISA analysis, Aβ peptides were first solubilized using 300μL of cold assay buffer composed of 50 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES), 5 mM ethylene-diamine-tetraacetic acid, 5 mM ethylene glycol tetraacetic acid, 30 mM NaCl, 1% Triton X-100, pH 7.4 with 0.1% bovine serum albumin (BSA), and then were further diluted 2-fold using assay buffer with 100μg/mL of heterophilic blocking reagent 1 (Scantibodies Laboratory, Inc. Santee, CA) added.

For ELISA measurement, anti-Aβ antibody 2G3 (specific for Aβ₄₀) or 21F12 (specific for Aβ₄₂) were used as capture antibodies, and ruthenium-labeled anti-Aβ 3D6 was used as detection antibody in a Meso Scale Discovery (MSD) format. The capture antibody was coated on an MSD plate for 16 to 24 h at a concentration of 10μg/mL. After coating, the plate was blocked using a cold phosphate buffered saline with 1% BSA. Aβ peptide standard curves were generated by diluting Aβ₄₀ and Aβ₄₂ in the assay buffer from 0.68 pg/mL to 40 ng/mL with a 3× dilution between neighboring points, plus the background control. Overall, 50μL of standards, diluted controls, and diluted samples was added to respective MSD plates for Aβ₄₀ and Aβ₄₂. After incubation for 2 h at 4°C with shaking, the plates were washed using a washing buffer of 1× TBS-Tween, pH 7.4, and 50μL of detection antibody at a concentration of 1μg/mL was added to each well, and the plates were further incubated at 4°C for 1 h with shaking. The plates were washed again using the washing buffer, and 150μL of freshly prepared 2× MSD Read Buffer (Meso Scale Discovery, Rockville, MD) was added to each well. The response data were collected on a Meso Scale Sector Imager SQ120 within 5 min from the addition of the Read Buffer. The data were analyzed using a 5-parameter logistic algorithm. The averaged recovery yield from the control samples was used to correct the Aβ peptide concentrations in the unknown samples.

Non-compartmental PK and PD analysis

Serum PK parameters for free LY2599666 were calculated using standard noncompartmental methods of analysis using Phoenix WinNonlin 6.4. The PK parameters calculated included maximum observed drug concentration (C_max), area under the concentration versus time curve (AUC), and half-life associated with the terminal rate constant (λz) in noncompartmental analysis (t_1/2). The actual venous blood sampling times were used in the calculation of PK parameters. For sampling time points with concentrations below the limit of quantification (BQL) that occurred before the first quantifiable concentration, the LY2599666 concentration was set to zero. All other postdose BLQ values were excluded from non-compartmental analyses.

Plasma PD biomarkers (total Aβ₄₀ and total Aβ₄₂) parameters were calculated using standard noncompartmental methods of analysis using Phoenix WinNonlin 6.4. The plasma PD parameters calculated included maximum Aβ₄₀ and Aβ₄₂ level (C_max), area under the PD responses versus time curve (AUC), and t_1/2.

Total (free + bound) LY2599666 concentration from available CSF samples from the 25 mg multiple dose cohort was quantified and the ratio of CSF total concentration to serum free concentration was computed for LY2599666.

Pharmacokinetic and pharmacodynamic model development

Nonlinear mixed-effect population analyses described in this section were conducted using NONMEM (version VII; ICON Development Solutions, Ellicott City, Maryland). The method used was ADVAN6 subroutine and the first-order conditional estimation with interaction. A two target-mediated drug disposition model was used to simultaneously fit free serum LY2599666 concentration and total plasma biomarkers (Aβ₄₀ and Aβ₄₂) from Part A data (Fig. 1). The key PK parameters estimated were first-order absorption rate constant from a depot (Ka), central elimination rate (Ke), and central volume (V1). Since human IV data are not available, bioavailability was assumed to be 1; hence CL/F equals Ke multiplying V1 (Ke*V1), and apparent volume (V1/F) equals V1. It is assumed that after binding, the complex of Aβ-LY2599666 was eliminated at the same rate (Ke) as LY2599666 mAb. The key PD parameters estimated were baseline plasma Aβ₄₀ (BS1) and Aβ₄₂ (BS2) level, the elimination rate of Aβ₄₀ (Kot1) and Aβ₄₂ (Kot2), and the binding affinity (Kd) of LY2599666 to biomarker (assumed to be the same between Aβ₄₀ and Aβ₄₂). The synthesis rate of Aβ₄₀ (Kin1) and Aβ₄₂ (Kin2) equal to BS1* Kot1 and BS2* Kot2, respectively.

Fig.1

Model scheme of two-target mediated drug disposition model for LY2599666.

Inter-patient variability was tested separately on each of the PK parameters using an exponential error structure (that is, log-normal distribution of individual parameter values) or additive error structure. Covariance between model parameters (CL and V1) in the PK model was examined by the application of omega blocks. A combined proportional and additive error structure was used to describe the residual error.

Selection of the most appropriate PK base model structure was based upon agreement between predicted and observed serum concentrations, lack of pattern (that is, randomness) in the weighted residuals versus the predicted values, magnitude of inter-patient variability, and significant decreases in the minimum objective function.

A visual predictive check (VPC) was performed on the model to ensure that the model maintained fidelity with the data upon which it was developed. The VPC approach entailed simulating PK and PD data with the developed model, taking into account variability in all parameters as given by the between-subject variability and residual error terms. The distribution of simulated concentrations was then compared to the observed distributions to ensure concordance. Simulated and observed distributions were compared by calculating the median, 5th, and 95th percentiles for each.

Additional, simulations were performed to predict PK and PD data after multiple doses of 25 mg SC QW of LY2599666. The distribution of simulated concentrations was then compared to the observed PK and PD data from Part B.

RESULTS

Demographics and sample disposition

In Part A, a total of 43 healthy subjects (17 Japanese and 26 non-Japanese subjects), 33 males and 10 females, between the ages of 23 and 65 years, were enrolled in the US and randomly assigned to receive single dose of the study treatment (32 received LY2599666; 11 received placebo). In Part B, a total of 7 patients with AD, 3 males and 4 females, between the ages of 58 and 76 years, were recruited in the US (3) and Japan (4) (5 LY2599666; 2 placebo). One of 7 patients did not complete the study and was withdrawn due to worsening of pre-existing AE of anemia after receiving 3 LY2599666 doses. Therefore, PKPD data from only 4 patients were included in evaluation of LY2599666.

A total of 512 serum samples from 32 healthy subjects treated with LY2599666 were collected in Part A single SC dose cohorts (10 mg to 200 mg). At 10 mg LY2599666, 112 of the 128 samples collected were BQL prior to the first quantifiable or after the last quantifiable sample. At 25 mg LY2599666, 50 of 128 samples collected were BQL prior to the first quantifiable or after the last quantifiable sample. At 100 mg, 10 of 128 samples collected were BQL prior to the first quantifiable or after the last quantifiable sample, and at 200 mg LY2599666, 26 of 128 serum samples were BQL prior to the first quantifiable or after the last quantifiable or had no result. A total of 162 serum samples were collected from 5 patients that received QW SC doses of 25 mg LY2599666 in Part B, in which 35 samples were BQL. Plasma total Aβ₄₀ and Aβ₄₂ samples were collected from 50 subjects in Parts A and B of the study that received either LY2599666 or placebo. Four CSF samples were collected from AD patients that received 25 mg LY2599666 QW and one sample was excluded due to blood contamination.

Non-compartmental PKPD analysis results

Following a single SC dose of LY2599666, mean concentration versus time profiles by treatment group for LY2599666 is shown in (Fig. 2). Since there were only 16 quantifiable serum samples collected following a single SC dose of 10 mg LY2599666, no data from 10 mg were included in the non-compartmental (NCA) analysis. A summary of NCA PK parameters is presented in Supplementary Table 1. It should be recognized that the terminal t_1/2 calculated by NCA PK analysis may not represent the true terminal t_1/2 due to impacts from the quantification limit. For example, most PK samples at 240 h were BQL following a single dose of 25 mg LY2599666, suggesting that quantifiable concentrations only represented less than three half-lives of its PK profile. Individual C_max and area under the concentration versus time curve from zero to infinity (AUC[0–∞]) values from the Japanese and non-Japanese populations were compared (not reported), suggesting no significant difference in exposures between the two populations.

Fig.2

Arithmetic mean LY2599666 plasma concentration versus time profiles following a single subcutaneous dose of LY2599666.

The plasma Aβ₄₀ and Aβ₄₂ response (geometric mean [% CV]) in healthy subjects who received placebo sustained at 213 (19) pg/mL and 52 (19) pg/mL with a range of 143 to 567 pg/mL, and 33 to 91 pg/mL, respectively. Following a single SC dose of LY2599666, both total plasma Aβ₄₀ and Aβ₄₂ showed dose-dependent increases (Fig. 3). Noncompartmental analyses were performed with plasma total Aβ₄₀ and Aβ₄₂ concentration versus time profiles (Supplementary Table 2).

Fig.3

Arithmetic mean Aβ plasma concentration versus time profiles following a single subcutaneous dose of LY2599666.

Supplementary Figure 1 presents all the PK data collected from 25 mg SC QW dose of LY2599666. The accumulation ratio was 1.27-fold based on geometric mean C_max and 1.39-fold based on geometric mean C_trough. The steady state appeared to be achieved around Day 22 based on C_trough.

The plasma Aβ₄₀ and Aβ₄₂ response (geometric mean [% CV]) in the 2 AD patients who received placebo for 12 weeks sustained at 247 (15) pg/mL and 50 (12) pg/mL with a range of 162 to 309 pg/mL and 41 to 66 pg/mL, respectively. Following the first dose of 25 mg LY2599666, total plasma Aβ₄₀ and Aβ₄₂ showed immediate increases, and maintained at high levels with steady state being achieved around Day 21 (Supplementary Figure 1).

Total LY2599666 concentrations in CSF are listed in (Table 1), with the time-matching serum free LY2599666 concentration from the same subject. The ratio between “total CSF” and “free serum” concentration was about 1% or higher. However, the molecule’s penetration capability may be overestimated by using “total CSF/free serum” ratio, as most LY2599666 in serum will bind to plasma Aβ₄₀ and Aβ₄₂, thus cannot be quantified by an antigen binding ELISA assay.

Table 1

LY2599666 cerebrospinal fluid concentrations for three patients administered 25 mg LY2599666 QW

Subject	Total CSF (ng/mL)	Free Plasma (ng/mL)	Total CSF/Free Plasma (%)
A	3.905	371	1.05
B	5.01	354.9	1.41
C	5.655	254.2	2.22
Geometric Mean (CV%)	4.80 (19)	322.27 (21)	1.49 (39)

CSF, cerebrospinal fluid; CV, coefficient of variation.

Population PKPD modeling results

Due to the prevalence of BQL samples in 10 mg/kg dose group, multiple BQL handling method was tested (M1, M3, M4) [4]. There were no significant difference in final parameter estimates and VPC across different BQL handling methods. Due to the longer computation time by using M3 and M4 methods, the final model was developed after omitting the BQL samples (M1 method).

The two-target mediated disposition model well described the free LY2599666 concentrations and total Aβ₄₀ and Aβ₄₂ levels of LY2599666 after a single SC administration. PK and PD parameters are summarized in Table 2. The population estimates were well characterized with adequate precision (≤18.9% relative standard error of estimate). The terminal t_1/2 was estimated as 7.3 days (175 h) at high doses when binding is saturated and PK becomes linear. Between-subject variability was small for central elimination rate constant (Ke) and central volume (V1), and high for the absorption rate (Ka). The interactions between Ke and V1 was moderate and also well estimated. The proportional and additive residual error for PK was relatively low (5.51% and 22.5 ng/mL, respectively).

Table 2

Pharmacokinetic and pharmacodynamic parameters for TMDD model

Parameter Description	Population Estimate (% RSE)	Between-subject Variability (% RSE)^a
Absorption rate Ka (1/h)	0.0225 (10.7%)	72.1% (25.3%)
Central elimination rate constant Ke (L/h)	0.00788 (5.5%)	16.5% (33.4%)
Central volume V1 (L)	12 (6.4%)	21.3% (37.8%)
Central to peripheral transfer rate constant K12	0.00262 (18.9%)
Peripheral to central transfer rate constant K21	0.00657 (11.2%)
Covariance for Ke and V1^b	–0.0199
Residual error for PK (proportional, %)	5.51% (15.5%)
Residual error for PK (additive) (nM)	0.150 (24.3%)
Plasma Aβ₄₀ baseline BS1 (nM)	0.0614 (3.3%)	17.7% (27.9%)
Plasma Aβ₄₂ baseline BS2 (nM)	0.0109 (2.4%)	14.9% (20.5%)
Plasma Aβ₄₀ elimination rate constant Kot1 (1/h)	2.46 (4.2%)
Plasma Aβ₄₂ elimination rate constant Kot2 (1/h)	1.66 (5.5%)	18.0% (57.1%)
Binding affinity Kd (nM)	0.0218 (7%)
Residual error for plasma Aβ₄₀ (proportional %)	4.95% (13.3%)
Residual error for plasma Aβ₄₀ (proportional %)	2.59% (10.7%)

Aβ, amyloid-β; CV, coefficient of variation; % RSE, relative standard error of estimation. ^a% CV = (SQRT(EXP(OMEGA(N))–1))*100%. ^bCovariance between ω ².

The estimated plasma baseline Aβ₄₀ and Aβ₄₂ was 265 pg/mL (0.0614 nM) and 49.2 pg/mL (0.0109 nM), respectively. The estimated plasma t_1/2 for Aβ₄₀ and Aβ₄₂ was 17 min and 25 min based on Kot1 and Kot2. The binding affinity (Kd) of LY2599666 to plasma Aβ₄₀ and Aβ₄₂ was assumed to be the same, which was estimated as 3.27 ng/mL (0.0218 nM). Low-moderate between-subject variability was also identified for plasma baseline Aβ₄₀ and Aβ₄₂, and plasma Aβ₄₂ elimination rate constant. All omega shrinkages were less than 30%. The proportional residual error for plasma Aβ was relatively low (<5%).

The final model was selected based upon goodness-of-fit plots demonstrating good comparability between predicted and observed, as well as demonstrating the randomness of the individual weighted residuals.

Population PK/PD model validation

A VPC was performed on the model to ensure that the model maintains fidelity with the data used to develop it. The VPC approach entails simulating PK, plasma Aβ₄₀ and Aβ₄₂ data 1000 times with the developed model, taking into account variability in all parameters, as given by the between-subject variability and residual error terms. The distributions of simulated concentrations were then compared to the actual concentration distributions to ensure concordance. The VPC plots are provided as Supplementary Figures 2–4.

The model built with Part A data was used to simulate 25 mg QW multiple dose PK, plasma Aβ₄₀ and Aβ₄₂ data, taking into account variability in all parameters, as given by the between-subject variability and residual error terms. The distributions of simulated concentrations were then compared to the actual individual data observed in Part B. Both total plasma Aβ₄₀ and Aβ₄₂ showed immediate increases following the first dose of 25 mg LY2599666, and are maintained at high levels with QW dosing frequency, with steady state being achieved around Day 21. As shown in Supplementary Figure 1, most observed PK and plasma Aβ₄₀ and Aβ₄₂ data from Part B distributed evenly on both sides of the predicted median from Part A model, and fell into 90% prediction interval, suggesting that the PK/PD model developed based on Part A data and predicted Part B data.

DISCUSSION

The target mediated drug disposition (TMDD) model has been utilized by pharmacometricians for more than two decades [5, 6] to describe the scenario when a pharmacological target impacts the apparent pharmacokinetics of a drug (typically a monoclonal antibody). About a decade ago, this TMDD model was expanded to describe scenarios when drugs bind to two targets, with a variety of assumptions [7 –9]. LY2599666 is expected to specifically bind to the mid-domain epitope of multiple soluble Aβ monomers, with similar affinity based on in vitro data (not reported). Multiple soluble Aβ isoforms present not only in the interstitial fluid of brain and CSF, but also in peripheral, for example, plasma [10]. Although numerous other isoforms exist in plasma, Aβ₄₀ and Aβ₄₂ are the primary isoforms present (∼80% – 90% for Aβ₄₀ and ∼5% – 10% for Aβ₄₂) [11] and therefore were collected in present Phase 1 study to confirm target engagement of LY2599666.

Baseline plasma Aβ₄₀ and Aβ₄₂ levels vary depending on the quantification assay and population [12]. The observed baseline plasma Aβ₄₀ (265 pg/mL, 0.0614 nM) and Aβ₄₂ (49.2 pg/mL, 0.0109 nM) levels in healthy subjects in this Phase I study generally aligned with published literature data (254.85 pg/mL for plasma Aβ₄₀ and 65.63 pg/mL for plasma Aβ₄₂) [13], confirming an approximate ratio of 5:1 between plasma Aβ₄₀ and plasma Aβ₄₂ level. The baseline Aβ₄₀ and Aβ₄₂ levels in patients, as what has been reported [14, 15], were not significantly different from healthy subjects. The estimated t_1/2 of plasma Aβ₄₀ (17 min) and Aβ₄₂ (25 min) from TMDD model were in the same range as published literature values [16, 17].

Nonlinear PK was observed for LY2599666 at doses less than 100 mg. This is because LY2599666 cannot be detected by antigen binding ELISA PK assay due to binding to plasma Aβ. At high dose levels, binding to free soluble Aβ would be saturated based on molar ratio between LY2599666 (post a single dose of 200 mg: >125 ng/mL, >1.86 nM at 672 h) and plasma Aβ level (baseline Aβ₄₀ + Aβ₄₂: 0.0723 nM), and thus PK became linear. The reported half-life of LY2599666 at high doses was consistent with the prolonged half-life reported for other PEGylated peptides [18, 19]. Accumulation was observed for plasma total Aβ after administration of LY2599666, which was because the t_1/2 of Aβ-LY2599666 complex was expected to be the same as LY2599666, and much longer than the t_1/2 of free Aβ. As expected, PK and PD results showed comparable t_1/2 between free LY2599666 at high doses (92.6 h–175 h), and total Aβ levels (117 h–147 h), and supports weekly dosing of LY2599666.

In addition, CSF samples were collected and the ratio between CSF exposure and serum exposure were much higher (1.49%) than published values (0.1–0.3%) from other mAb [20 –22]. This is likely due to different composition between a free LY2599666 in serum and a total LY2599666 (free LY2599666 + Aβ-LY2599666 complex) level. The ratio between projected total LY in serum based on TMDD model and observed total LY2599666 in CSF was approximately 0.5%, thus aligned with current understanding of penetration ratio.

In summary, about one week half-life was shown for LY2599666, supporting weekly SC dosing. Strong target engagement was demonstrated with a dose dependent accumulation of total plasma Aβ₄₀ and Aβ₄₂. TMDD model developed based on SAD healthy subject data predicted target engagement in AD patients in weekly multiple dose regimen.

Footnotes

ACKNOWLEDGMENTS

The authors wish to acknowledge the investigators and subjects who participated in the clinical study Authors’ disclosures available online

The supplementary material is available in the electronic version of this article: .

References

Hardy

(1996) Molecular genetics of Alzheimer's disease. Acta Neurol Scand Suppl 165, 13–17.

Selkoe

(1991) The molecular pathology of Alzheimer's disease. Neuron 6, 487–498.

Price

, Tanzi

, Borchelt

, Sisodia

(1998) Alzheimer's disease: Genetic studies and transgenic models. Annu Rev Genet 32, 461–493.

Bergstrand

, Karlsson

(2009) Handling data below the limit of quantification in mixed effect models. AAPS J 11, 371–380.

Levy

, Mager

, Cheung

, Jusko

(2003) Comparative pharmacokinetics of coumarin anticoagulants L: Physiologic modeling of S-warfarin in rats and pharmaco-logic target-mediated warfarin disposition in man. J Pharm Sci 92, 985–994.

Levy

(1994) Pharmacologic target-mediated drug disposition. Clin Pharmacol Ther 56, 248–252.

Gibiansky

, Gibiansky

(2010) Target-mediated drug disposition model for drugs that bind to more than one target. J Pharmacokinet Pharmacodyn 37, 323–346.

Gibiansky

, Gibiansky

, Kakkar

, Ma

(2008) Approximations of the target-mediated drug disposition model and identifiability of model parameters. J Pharmacokinet Phar-macodyn 35, 573–591.

Mager

, Krzyzanski

(2005) Quasi-equilibrium phar-macokinetic model for drugs exhibiting target-mediated drug disposition. Pharm Res 22, 1589–1596.

10.

Wang

, Gu

, Masters

, Wang

(2017) A systemic view of Alzheimer disease - insights from amyloid-beta metabolism beyond the brain. Nat Rev Neurol 13, 703.

11.

Murphy

, LeVine

3rd (2010) Alzheimer's disease and the amyloid-beta peptide. J Alzheimers Dis 19, 311–323.

12.

Shanthi

, Krishnan

, Rani

(2015) A systematic review and meta-analysis of plasma amyloid 1-42 and tau as biomarkers for Alzheimer's disease. SAGE Open Med 3, 2050312115598250.

13.

Poljak

, Crawford

, Smythe

, Brodaty

, Slavin

, Kochan

, Trollor

, Wen

, Mather

, Assareh

, Ng

, Sachdev

(2016) The relationship between plasma Abeta levels, cognitive function and brain volumetrics: Sydney Memory and Ageing Study. Curr Alzheimer Res 13, 243–255.

14.

Rembach

, Watt

, Wilson

, Villemagne

, Burnham

, Ellis

, Maruff

, Ames

, Rowe

, Macaulay

, Bush

, Martins

, Masters

, Doecke

, AIBL Research Group (2014) Plasma amyloid-beta levels are significantly associated with a transition toward Alzheimer's disease as measured by cognitive decline and change in neocortical amyloid burden. J Alzheimers Dis 40, 95–104.

15.

Olsson

, Lautner

, Andreasson

, Ohrfelt

, Portelius

, Bjerke

, Holtta

, Rosen

, Olsson

, Strobel

, Wu

, Dakin

, Petzold

, Blennow

, Zetterberg

(2016) CSF and blood biomarkers for the diagnosis of Alzheimer's disease: A systematic review and meta-analysis. Lancet Neurol 15, 673–684.

16.

Golde

, Das

, Levites

(2009) Quantitative and mechanistic studies of Abeta immunotherapy. CNS Neurol Disord Drug Targets 8, 31–49.

17.

Ovod

, Ramsey

, Mawuenyega

, Bollinger

, Hicks

, Schneider

, Sullivan

, Paumier

, Holtzman

, Morris

, Benzinger

, Fagan

, Patterson

, Bateman

(2017) Amyloid beta concentrations and stable isotope labeling kinetics of human plasma specific to central nervous system amyloidosis. Alzheimers Dement 13, 841–849.

18.

Caliceti

, Veronese

(2003) Pharmacokinetic and biodistribution properties of poly(ethylene glycol)-protein conjugates. Adv Drug Deliv Rev 55, 1261–1277.

19.

Harris

, Chess

(2003) Effect of pegylation on pharmaceuticals. Nat Rev Drug Discov 2, 214–221.

20.

Frolich

, Kornhuber

, Ihl

, Fritze

, Maurer

, Riederer

(1991) Integrity of the blood-CSF barrier in dementia of Alzheimer type: CSF/serum ratios of albumin and IgG. Eur Arch Psychiatry Clin Neurosci 240, 363–366.

21.

Pestalozzi

, Brignoli

(2000) Trastuzumab in CSF. J Clin Oncol 18, 2349–2351.

22.

Siemers

, Friedrich

, Dean

, Gonzales

, Farlow

, Paul

, Demattos

(2010) Safety and changes in plasma and cerebrospinal fluid amyloid beta after a single administration of an amyloid beta monoclonal antibody in subjects with Alzheimer disease. Clin Neuropharmacol 33, 67–73.