Biomarker Exposure-Response Analysis in Mild-To-Moderate Alzheimer’s Disease Trials of Bapineuzumab

Abstract

Background: Bapineuzumab, an anti-amyloid monoclonal antibody, was evaluated as a candidate for immunotherapy in mild-to-moderate Alzheimer’s disease (AD) patients.

Objective: To assess the treatment effect of bapineuzumab therapy on disease-relevant biomarkers in patients with mild-to-moderate AD, using exposure-response modeling.

Methods: Biomarker data from two Phase III studies were combined to model the impact of bapineuzumab exposure on week-71 change from baseline in brain amyloid burden by ¹¹C-labeled Pittsburgh compound B (PiB) PET imaging (global cortical average of the Standardized Uptake Value ratio values), cerebrospinal fluid (CSF) phosphorylated (p)-tau concentrations, and brain volumetrics (brain boundary shift integral) by magnetic resonance imaging. Bapineuzumab or placebo was administered as a 1-hour intravenous infusion every 13 weeks for 78 weeks. Pharmacokinetic/pharmacodynamic modeling helped determine the most appropriate exposure-response model and estimate the impact of disease-relevant covariates (baseline biomarker value, APOE^*E4 allele copy number, and baseline disease status as measured by Mini-Mental State Examination score) on the three biomarkers.

Results: Linear exposure-response relationships with negative and significant slope terms were observed for PiB PET and CSF p-tau concentration. Baseline biomarker value and APOE^*E4 carrier status were significant covariates for both biomarkers. No exposure-response relationship on brain boundary shift integral was detected.

Conclusions: Bapineuzumab treatment induced exposure-dependent reductions in brain amyloid burden. Effects on CSF p-tau concentrations were significant only in APOE*E4 carriers. No apparent influence of bapineuzumab exposure on brain volume could be demonstrated.

Keywords

Amyloid PET bapineuzumab brain volume cerebrospinal fluid disease severity exposure-response modeling p-tau

INTRODUCTION

Two hallmark pathological features of Alzheimer’s disease (AD) are neuritic plaques and neurofibrillary tangles [1]. Amyloid plaques consist of aggregates of amyloid-β (Aβ) peptide, particularly the 42-amino acid isoform (Aβ₄₂) [2, 3]. Abnormal Aβ₄₂ processing in AD leads to an increased brain amyloid burden that serves as a biomarker, which can be detected by positron emission tomography (PET) imaging [4]. AD is also accompanied by an elevation in cerebrospinal fluid (CSF) total tau (t-tau) and phosphorylated tau181 (p-tau) proteins, which are biomarkers of neuronal injury and neurofibrillary tangle formation [5 –10]. Greater brain atrophy rate reflecting increased neuronal loss and neurodegeneration occurs in patients with AD as compared with age and sex-matched healthy controls [11 –14].

Bapineuzumab, a recombinant humanized anti-Aβ peptide monoclonal antibody, directed against the N-terminus (amino acids 1 to 5) of human Aβ_1 - 40/42 peptides was the first candidate for immunotherapy in AD. It is thought to act in multiple ways, including direct capture and neutralization of soluble Aβ monomers and oligomers, and disruption and clearance of parenchymal and vascular Aβ deposits by either direct dissolution of fibrillar material or Fc-mediated phagocytosis (principally via microglia) [15 –17]. Evidence of reduced brain amyloid burden by¹¹C labeled Pittsburgh compound B (PiB) PET imaging and CSF p-tau levels in mild-to-moderate AD patients in Phase II trials [18, 19] enabled advancement to Phase III evaluation [20].

In the present work, data from two double-blind, randomized, placebo-controlled Phase III studies [20] in patients with mild-to-moderate AD were combined to model the pharmacodynamics (PD) of intravenous bapineuzumab and to estimate the impact of bapineuzumab exposure on the change from baseline to week-71 in brain amyloid burden, CSF p-tau concentrations, and brain volume. In addition, possible covariate effects of baseline value of biomarkers, apolipoprotein (APOE*E4) allele copy number, and baseline disease status on biomarkers were assessed. The effects of disease progression, and possible placebo effects were also evaluated for these biomarkers.

MATERIALS AND METHODS

Clinical data

Two Phase III clinical trials (NCT00575055, NCT00574132), one each in APOE*E4 carriers and non-carriers, were conducted in patients with mild-to-moderate AD dementia [20]. Enrollment criteria for the bapineuzumab biomarker substudies were the same as for the main studies described previously [20]. Briefly, eligible patients aged 50 to 88 years inclusive, met the criteria for probable AD of the National Institute of Neurological and Communicative Disorders and Stroke and the Alzheimer’s disease and Related Disorders Association [21], had a score of 16–26 on the Mini–Mental State Examination (MMSE) [22], and a score≤4 on the Rosen modified Hachinski Ischemic scale [23]. Patients with neurological disease other than AD were excluded. The institutional review board for each site approved the study, and each patient (or legally authorized representative) gave written informed consent before enrollment.

Clinical study design and treatment

Bapineuzumab was administered as 1-hour intravenous infusions every 13 weeks for a 78-week study period. In the APOE*E4 carrier study, 1,121 patients were randomized in a 3:2 ratio to 0.5 mg/kg or placebo. In the APOE*E4 noncarrier study, 1,331 patients were randomized in a 1:1:1:2 ratio to 0.5 mg/kg, 1.0 mg/kg, 2.0 mg/kg of bapineuzumab, or placebo. The 2.0 mg/kg dose was discontinued early in the trial due to a high rate of clinically symptomatic amyloid-related imaging abnormalities with effusion or edema (ARIA-E). Patients randomized to the 2.0 mg/kg dose level continued in the study at the 1.0 mg/kg dose level. Subsets of patients underwent amyloid PET, CSF, and magnetic resonance imaging (MRI) assessments.

The co-primary endpoints of both studies were clinical measures of efficacy of bapineuzumab versus placebo on the change from baseline to week-78. For both studies (NCT00575055, NCT00574132), key secondary objectives were to evaluate the effect of bapineuzumab on change from baseline to week-71 of three disease relevant biomarkers: Brain amyloid burden (PiB PET), CSF p-tau concentrations, and brain volume.

The PiB-PET global cortical average was calculated as the average standardized uptake value ratio (GCA SUVr) from five brain regions of interest (ROI) known to accumulate amyloid in AD (anterior cingulate cortex, posterior cingulate cortex or precuneus, frontal cortex, lateral temporal cortex, and parietal cortex) [24]. Patient CSF p-tau samples were obtained by lumbar puncture and were analyzed using a sandwich ELISA method (Innogenetics, Fujirebio) [25]. Brain volume changes were assessed via the brain boundary-shift-integral method (BBSI), as measured by volumetric MRI [26].

PiB PET GCA SUVr values, CSF p-tau concentrations, and brain volume were assessed predose, (i.e., before the first infusion on day 1), or at screening and at weeks 45/71 (PET), week-71, and weeks 19/45/71, respectively. Baseline and week-71 measures were utilized in this analysis. The analysis population included patients that had both baseline and week-71 biomarker measurements (for a given biomarkers), and those who received at least 5 infusions, with 1 of the infusions at week 65 (i.e., last dose). Patients who tested negative for amyloid at baseline (i.e., SUVr values <1.35) were also included in the PiB PET analysis, and baseline SUVr was investigated as a potential covariate. Patient characteristics are reported in Table 1. The demographic characteristics and baseline disease severity in the three biomarkers sub-populations were generally similar in the placebo and bapineuzumab groups (Table 2).

Pharmacokinetic parameters

Individual exposure metrics considered for the exposure-response analyses were obtained from previous population pharmacokinetic (PK) modeling of bapineuzumab serum concentration data [27], namely: Steady-state area under the concentration-time curve (AUC) calculated as average dose divided by post hoc clearance, steady-state trough concentration (CPSS), and week-71 concentration (CP71). In the CSF p-tau analysis, observed CSF bapineuzumab concentration at week 71 (CS71; available only in the CSF substudy) was also considered.

Pharmacokinetic/pharmacodynamic modeling of exposure-response relationship

The exposure-response analyses were planned before database lock of the two Phase III trials occurred. An analysis plan, further described below, was pre-specified for each biomarker and was consistently applied for the data analysis. For each biomarker, the analysis consisted of a preliminary exploration of the response variable (change from baseline to week-71), to evaluate which exposure metric correlated best with response and to visually assess the shape of the exposure-response relationship. Subsequently, PK/PD modeling was performed to estimate the parameters and capture the shape of the exposure-response relationship. A sequential PK/PD modeling approach was adopted, i.e., individual PK parameters were first fixed at their individual Bayesian estimates obtained from the population PK analysis [27] and were then used to drive the exposure-response PK/PD model. The model-building procedure entailed the following steps: ‘base model’, ‘covariate model’, and ‘final model’ building.

In the ‘base-model’ building step several direct-effect PK/PD models, including linear, maximum effect (E_max), or sigmoid (E_max model with Hill coefficient) models, were used to link individual predicted exposure to biomarker change, without introducing covariates. The most suitable functional form was determined through parameter significance and goodness-of-fit, and was established as the ‘base model’.

Next, the ‘covariate model’ building step was performed, whereby the influence of the covariates on both the placebo and bapineuzumab parameters in the ‘base model’ was quantitatively assessed. Baseline biomarker values were assessed as continuous covariates centered on their respective means; in the BBSI analysis, baseline whole brain volume (WBV), was tested as a covariate. APOE*E4 allele copy number was tested as a categorical covariate, or as carriers versus noncarriers (if no statistically significant difference was found between homozygote and heterozygote subjects). Baseline disease status was assessed as either ‘mild’ AD (baseline MMSE score 21–26 inclusive) or ‘moderate’ AD (baseline MMSE score 16–20 inclusive). Each covariate was tested separately.

In order to obtain the ‘final model’, all statistically significant covariate contributions derived from the ‘covariate model’ building step were implemented simultaneously on top of the base model, and non-significant contributions were removed.

In all steps, precision and significance of the parameter estimates (using a 2-sided threshold of α= 0.05), as well as goodness of fit, (i.e., sum of squared residuals) were used to assess the functional form of the exposure-response model and the need for the various covariate contributions.

The results were presented in graphical format as the fitted exposure-response curve with the associated 95% prediction intervals in graphical format. The prediction intervals captured the influence of parameter uncertainty in the model projections, as well as random variation in the data. For a given exposure level, the prediction interval provided the range in which a new observation (i.e., biomarker change from baseline) was expected to lie with 95% probability. R software (version 2.15.0) was used for dataset exploration, visualization, model identification, and construction of diagnostic plots [28].

RESULTS

PiB PET analysis

Data exploration using summary statistics showed decreased amyloid burden in the bapineuzumab-treated groups versus placebo, i.e., less accumulation (0.5 mg/kg) or reduction (1.0 and 2.0 mg/kg) from baseline were observed at week 71. Baseline SUVr values in the placebo and in the bapineuzumab groups were comparable (Table 2). Among the three serum exposure metrics evaluated, bapineuzumab AUC was the most correlated with the decrease in GCA SUVr values from baseline.

A linear model described the data appropriately, and was thus used as the ‘base model’. A preliminary fit of the base model showed a negative and significant slope term (p < 0.01). In the ‘covariate model’ building step, the baseline biomarker effect on the model slope was significant (p < 0.05). Moreover, APOE*E4 carrier status was a significant covariate on the intercept when testing noncarriers versus carriers: The difference in intercept (carriers minus noncarriers) was significantly different from zero (p < 0.001). No significant difference between homozygote and heterozygote carriers was found. A statistically significant effect of baseline disease status on the model parameters was not found.

The final model is given by the following equations: $\begin{matrix} {CFB}_{i} = E_{0, i} + E_{D, i} \times {AUC}_{i} + ɛ_{i} \\ E_{0, i} = C_{i} \times E_{0, C} E_{D, i} = E_{D} + E_{D, BAS} \\ \times ({BAS}_{i} - {BAS}_{avg}) \end{matrix}$ where, for the i-th subject, CFB_i = the change from baseline of biomarker value at week-71, AUC_i = the value of bapineuzumab AUC, ɛ_i = the additive contribution of residual variability, BAS_i = the baseline biomarker value, BAS_avg = the average baseline biomarker value across all patients (1.948), E_0,i = intercept value, equal to ‘0’ for APOE*E4 non-carriers (C_i = 0) or E_0,C for carriers (C_i = 1), E_D,i = slope parameter, E_D = the typical value of the slope parameter, and E_D,BAS = the slope contribution associated with the baseline GCA SUVr value. In the ‘final model’ building step, the intercept parameter for noncarriers was not significantly different from zero (0.0148, p = 0.58), and was therefore excluded from the ‘final model’. Moreover, the negative slope was statistically significant. Parameter estimates of the final PiB PET GCA SUVr model are reported in Table 3.

The ‘final model’ was also evaluated both at the average exposure level in the bapineuzumab-treated (AUC_avg = 259.2μg.d/mL) and at zero exposure (i.e., in the placebo group). The baseline biomarker value was set to the average value, BAS_avg. The difference between the two changes from baseline (equal to E_D×AUC_avg) was then evaluated to obtain a measure of the magnitude of effect in the bapineuzumab-treated versus the placebo group (–0.0498 [CV, 32% ]). This corresponds to a relative percentage biomarker reduction from baseline (bapineuzumab minus placebo) of approximately –3% (negative values favor bapineuzumab). This percentage reduction was calculated as the percentage biomarker reduction obtained with bapineuzumab minus the reduction obtained with placebo, and is equivalent (after substituting the model equations) to (E_D×AUC_avg / BAS_avg)×100. The fitted exposure-response curve and its 95% prediction interval described the dataset satisfactorily in APOE*E4 carriers and noncarriers (Fig. 1).

CSF p-tau concentration analysis

A greater reduction in CSF p-tau concentration from baseline at week-71 in the bapineuzumab groups was observed versus placebo. Baseline CSF p-tau concentrations were comparable in the placebo and bapineuzumab groups (Table 2). Among the four exposure metrics, CPSS (trough serum concentration at steady-state) was most correlated with biomarker change.

A linear function was deemed an appropriate base model to describe the data and provided a negative and significant slope term (p < 0.01).

The ‘covariate model’ building step showed significant effects of baseline CSF p-tau concentration on intercept (p < 0.0001). APOE*E4 carrier status was a significant covariate on both intercept and slope when testing noncarriers versus carriers: Both intercept difference (p < 0.01) and slope difference (p < 0.01) (i.e., carriers minus noncarriers) were significantly different from zero. No significant difference between homozygote and heterozygote carriers was found. Additionally, the baseline disease status was a significant covariate of intercept, i.e., intercept difference (moderate AD minus mild AD) was significantly different from zero (p < 0.01).

The final model is given by the following equations: $\begin{matrix} {CFB}_{i} = E_{0, i} + E_{D, i} \times {CPSS}_{i} + ɛ_{i} \\ E_{0, i} = E_{0} + E_{0, BAS} \times ({BAS}_{i} - {BAS}_{avg}) \\ + C_{i} \times Δ_{C} + {AD}_{0, i} \times Δ_{AD} E_{D, i} = C_{i} \times E_{D, C} \end{matrix}$ where, for the i-th subject, CPSS_i = the value of CPSS, BAS_avg = 107.248 pg/mL, E₀ = base intercept value, E_0,BAS = the intercept contribution associated with the baseline CSF p-tau concentration; patients who were APOE*E4 carriers (C_i = 1) and/or had moderate AD at baseline (AD_0,i = 1) contributed to the typical intercept with additive effects Δ_C and Δ_AD, respectively (AD_0,i = 0 for patients with mild AD at baseline); E_D_,C = the slope associated with APOE*E4 carriers. The slope associated with noncarriers, was not significantly different from zero (–2.302, p = 0.12), and was therefore excluded from the final model. Parameter estimates of the final CSF p-tau model are reported in Table 3.

The final model was also evaluated for APOE*E4 carriers both at the average exposure level in the bapineuzumab-treated (CPSS_C,avg = 0.792μg/mL), and at zero exposure (i.e., in the placebo group). The baseline biomarker value was set to the average value in APOE*E4 carriers, BAS_C,avg = 112.677 pg/mL. The difference between the two changes from baseline (equal to E_D_,C×CPSS_C,avg) was then evaluated to obtain a measure of the magnitude of effect in the bapineuzumab-treated versus the placebo group. The magnitude of the effect and percentage coefficient of variation was –5.822 (27%). This corresponds to a relative percentage biomarker reduction from baseline (bapineuzumab minus placebo) of approximately –5% (negative values favor bapineuzumab). This percentage reduction was calculated as the percentage biomarker reduction obtained with bapineuzumab minus the reduction obtained with placebo, and is equivalent (after substituting the model equations) to (E_D_,C×CPSS_C,avg/BAS_C,avg)×100. The effect magnitude for APOE*E4 noncarriers was zero because the associated exposure-response relationship was flat. The fitted exposure-response curve and its 95% prediction interval described the dataset satisfactorily in all four subpopulations (i.e., APOE*E4 carriers or noncarriers, and patients with mild AD or moderate AD) (Fig. 2).

BBSI analysis

Summary statistics suggested similar distributions of BBSI in the bapineuzumab and placebo groups. Baseline WBV values in the placebo and bapineuzumab groups were comparable (Table 2).

Since no exposure-response relationship could be shown, the subsequent modeling steps were undertaken for an intercept-only model (i.e., a model without effect of drug exposure), using both placebo and bapineuzumab data. The ‘covariate model’ building step showed significant effects on the intercept parameter of baseline WBV (p < 0.01) and APOE*E4 carrier status for carriers relative to noncarriers (p < 0.01). No significant difference between homozygote and heterozygote carriers was found. Additionally, the intercept difference evaluated as moderate AD minus mild AD was significant (p < 0.0001).

In the ‘final model’ building step, where all significant covariates were implemented simultaneously on top of the base model, the APOE*E4 effect on intercept was ruled out in terms of statistical significance (possibly because of the presence of the other covariates in the model), and was therefore excluded from the final model. The final model, is given by the following equations: $\begin{matrix} {CFB}_{i} = E_{0, i} + ɛ_{i} \\ E_{0, i} = E_{0} + E_{0, BAS} \times ({BAS}_{i} - {BAS}_{avg}) + {AD}_{0} \times Δ_{AD} \end{matrix}$ where BAS_avg = 1011.55 mL and E_0,BAS = the intercept contribution associated with the baseline WBV. The parameter estimates, percentage coefficients of variation, and p-values are shown in Table 3. The fitted exposure-response curve and its 95% prediction interval described the dataset satisfactorily both in patients with mild and moderate baseline AD (Fig. 3).

DISCUSSION

Modeling exposure-response relationships of biomarker changes in AD is useful not only to assess treatment effects and inform dose selection, but also to guide patient selection for clinical trials designed to test new therapeutic candidates. The findings can help reduce inter- and intra-patient variability and lead to predictive and prognostic enrichment of those patients likely to respond to anti-amyloid therapy [29]. The bapineuzumab Phase III trials were designed to evaluate disease modification based on a combination of clinical and biomarker evidence. Key biomarkers were selected based on results from two Phase II studies [18, 30], and were used to investigate the disease modification mechanism of bapineuzumab therapy. Exposure-response modeling of PiB PET GCA SUVr values and CSF p-tau concentrations in patients with mild-to-moderate AD showed a linear relationship between bapineuzumab exposure and change from baseline to week-71.

In the PiB PET analysis, the negative and significant slope indicates that bapineuzumab reduced brain fibrillar amyloid accumulation, as compared to placebo. The baseline SUVr value was a significant covariate on the slope of the exposure-response relationship, i.e., higher values of baseline biomarkers were associated with larger biomarker decreases from baseline. The treatment effect of bapineuzumab was significant in both APOE*E4 carriers and non-carriers. Although the slope for the non-carrier group was numerically smaller in magnitude than for the carrier group, the difference between slopes was not significant. APOE*E4 carriers status was a significant covariate of the intercept parameter. The intercept of APOE*E4 carriers was significantly greater than zero, which suggested an amyloid accumulation over the treatment period consistent with disease progression in the placebo group. However, the intercept parameter was not significantly different from zero in APOE*E4 noncarriers, i.e., no disease progression effect was noted in the placebo group.

The treatment effect of bapineuzumab on CSF p-tau concentrations was significant only for APOE*E4 carriers. In the non-carrier group, the slope parameter estimated during model building was negative but not statistically significant. From a biological perspective, the relatively smaller and not statistically significant drug effect observed in APOE*E4 non-carriers suggests that larger exposures may be needed to decrease CSF p-tau in this group of patients (change from placebo was significant in the 1.0 mg/kg dose group but not in the 0.5 mg/kg group [31]). Moreover, exposures in the bapineuzumab Phase III studies were restricted by the dose limitations that were introduced because of ARIA-E. It should be noted that the design of the two Phase III studies, whereby one study enrolled only carriers while the other enrolled only non-carriers, does not rule out the possibility that these findings may reflect a study effect, as opposed to a biological effect of APOE*E4 genotype itself. Baseline p-tau concentration was a significant covariate of the intercept parameter as higher baseline p-tau concentrations were associated with larger p-tau decreases from baseline. APOE*E4 carrier status and baseline disease status were statistically significant covariates on intercept, i.e., carriers showed larger increase in p-tau concentrations than non-carriers, and patients with moderate AD at baseline showed greater CSF p-tau reductions than patients with mild AD at baseline. However, the large negative intercept for APOE*E4 non-carriers with moderate baseline AD (–7.685 pg/mL, Table 3) may be caused by an extreme point around –50 pg/mL in the placebo group. The intercept estimates for the other 3 groups are generally consistent with little or no disease progression nor improvement on CSF p-tau.

In the BBSI dataset, there was no evidence of an exposure-response relationship, as the distribution of response in the bapineuzumab and placebo groups were similar. Baseline WBV and baseline disease status were significant covariates in the intercept-only model. In particular, a greater degree of brain atrophy was observed in patients with moderate AD at baseline and in patients with a higher baseline WBV. In each biomarker analysis, the parameters of the final model were estimated with good precision. The fitted model described the dataset satisfactorily in terms of visual predictive checks. The distribution of residuals and the goodness- of-fit plots confirmed model suitability.

A cascade of biomarkers abnormalities has been proposed for the progression of AD, such that Aβ deposition precedes tau-mediated neuronal injury, followed by worsening on neuroimaging parameters, and then cognitive and functional measures [32]. At a given time point in this sequence, each individual presents with a pattern of pathophysiological changes associated with cognitive impairment. In an earlier study, a 86% agreement between amyloid PET and CSF p-tau biomarkers was demonstrated which could explain the similar results of linear exposure-response relationship for these two biomarkers [33]. However, differences in the impact of covariates on this exposure-response relationship were observed in the current analysis. These differences could be more consistent with a multifactorial model of AD [34 –37].

In the analyses of the two Phase III AD studies, the statistical significance of treatment effect on amyloid burden (PiB PET) and CSF p-tau did not translate into changes of the clinical endpoints [20], possibly because the magnitude of the biomarkers change was insufficient to elicit clinical benefit because of insufficient exposures (due to the dose limitation), and/or because a population with mild-to-moderate AD is too late in the AD spectrum to benefit from bapineuzumab treatment. Additionally, no differences in CSF Aβ_x - 40, Aβ_x - 42 and total tau between bapineuzumab and placebo were observed, although analysis of the CSF p-tau results suggests downstream effects from the Aβ target [31]. Absence of an apparent exposure-response on BBSI as well as the increased ventricular BBSI rates in the bapineuzumab-treated versus placebo may be explained by confounding factors such as removal of amyloid plaques, reduction of inflammation, and cerebral fluid shifts [30 , 39].

One limitation of this analysis was that about 36% of enrolled non-carriers were actually amyloid negative (this was partially taken into account in the PiB PET model by introducing the baseline SUVr as a covariate). In addition, only week-71 change from baseline (i.e., the key biomarker secondary endpoints of the two studies) rather than all available time points (4 MRIs, 3 PET scans) were included in the exposure-response analysis. Additionally, exploratory biomarkers (e.g., CSF Aβ, CSF tau, ventricular boundary shift integrals, etc.) were not modeled. Moreover, the different exposure metrics (AUC, CPSS, CP71, CS71) were highly correlated among themselves, hence the exposure-response modeling results (obtained using AUC for PET and CPSS for CSF p-tau) are expected to be similar with other exposure metrics as well. Finally, a simple linear function was used to model the impact of baseline biomarker values on intercept and slope parameters, which differs from standard approaches such as exponential or power covariate models (commonly used, for example, in population PK analysis).

Bapineuzumab is capable of inducing modifications in brain amyloid burden and CSF p-tau concentrations in patients with mild-to-moderate AD. However, no apparent influence of bapineuzumab exposure on brain volume could be demonstrated. Using bapineuzumab to target patients with genetic risk factors (such as being APOE*E4 carrier) and at an earlier disease stage, before overt cognitive deterioration but after brain amyloid accumulation has already started (e.g., patients with mild cognitive impairment), may potentially reveal a relationship between biomarker changes and slowing of cognitive decline.

Footnotes

ACKNOWLEDGMENTS

The study was funded by Janssen Alzheimer Immunotherapy Research & Development, LLC, and Pfizer Inc. We thank the study participants, without whom this study would never have been accomplished, and the investigators for their participation in this study. Medical writing support was provided by Ashwini Patil (SIRO Clinpharm Pvt. Ltd.), and Bradford Challis (Janssen Research & Development, LLC) provided additional editorial support and review of this manuscript.

This study was sponsored by Pfizer Inc. and Janssen Alzheimer Immunotherapy Research & Development, LLC. Medical writing support was provided by Ashwini Patil at SIRO Clinpharm Pvt. Ltd. and was funded by Janssen Research and Development, LLC. The sponsors also provided a formal review of this manuscript.

Part of the data was presented at the American Conference on Pharmacometrics (ACoP), October 12–15, 2014, Las Vegas, NV.

Authors’ disclosures available online ().

References

Bloom

(2014) Amyloid-beta and tau: The trigger and bullet in Alzheimer disease pathogenesis. JAMA Neurol 71, 505–508.

Hardy

, Selkoe

(2002) The amyloid hypothesis of Alzheimer’s disease: Progress and problems on the road to therapeutics. Science 297, 353–356.

Hong

, Quintero-Monzon

, Ostaszewski

, Podlisny

, Cavanaugh

, Yang

, Holtzman

, Cirrito

, Selkoe

(2011) Dynamic analysis of amyloid beta-protein in behaving mice reveals opposing changes in ISF versus parenchymal Abeta during age-related plaque formation. J Neurosci 31, 15861–15869.

Klunk

, Engler

, Nordberg

, Wang

, Blomqvist

, Holt

, Bergström

, Savitcheva

, Huang

, Estrada

, Ausen

, Debnath

, Barletta

, Price

, Sandell

, Lopresti

, Wall

, Koivisto

, Antoni

, Mathis

, Langstrom

(2004) Imaging brain amyloid in Alzheimer’s disease with Pittsburgh Compound-B. Ann Neurol 55, 306–319.

Buerger

, Ewers

, Pirttilä

, Zinkowski

, Alafuzoff

, Teipel

, DeBernardis

, Kerkman

, McCulloch

, Soininen

, Hampel

(2006) CSF phosphorylated tau protein correlates with neocortical neurofibrillary pathology in Alzheimer’s disease. Brain 129, 3035–3041.

Hampel

, Burger

, Teipel

, Bokde

, Zetterberg

, Blennow

(2008) Core candidate neurochemical and imaging biomarkers of Alzheimer’s disease. Alzheimers Dement 4, 38–48.

Hansson

, Zetterberg

, Buchhave

, Londos

, Blennow

, Minthon

(2006) Association between CSF biomarkers and incipient Alzheimer’s disease in patients with mild cognitive impairment: A follow-up study. Lancet Neurol 5, 228–234.

Jack

Jr , Wiste

, Vemuri

, Weigand

, Senjem

, Zeng

, Bernstein

, Gunter

, Pankratz

, Aisen

, Weiner

, Petersen

, Shaw

, Trojanowski

, Knopman

(2010) Brain beta-amyloid measures and magnetic resonance imaging atrophy both predict time-to-progression from mild cognitive impairment to Alzheimer’s disease. Brain 133, 3336–3348.

Shaw

, Vanderstichele

, Knapik-Czajka

, Clark

, Aisen

, Petersen

, Blennow

, Soares

, Simon

, Lewczuk

, Dean

, Siemers

, Potter

, Lee

, Trojanowski

(2009) Cerebrospinal fluid biomarker signature in Alzheimer’s disease neuroimaging initiative subjects. Ann Neurol 65, 403–413.

10.

Tapiola

, Alafuzoff

, Herukka

, Parkkinen

, Hartikainen

, Soininen

, Pirttila

(2009) Cerebrospinal fluid beta-amyloid 42 and tau proteins as biomarkers of Alzheimer-type pathologic changes in the brain. Arch Neurol 66, 382–389.

11.

Barnes

, Bartlett

, van de Pol

, Loy

, Scahill

, Frost

, Thompson

, Fox

(Aging) (2009) A meta-analysis of hippocampal atrophy rates in Alzheimer’s disease. Neurobiol 30, 1711–1723.

12.

Fox

, Crum

, Scahill

, Stevens

, Janssen

, Rossor

(2001) Imaging of onset and progression of Alzheimer’s disease with voxel-compression mapping of serial magnetic resonance images. Lancet 358, 201–205.

13.

Frisoni

, Fox

, Jack

Jr , Scheltens

, Thompson

(2010) The clinical use of structural MRI in Alzheimer disease. Nat Rev Neurol 6, 67–77.

14.

Schott

, Price

, Frost

, Whitwell

, Rossor

, Fox

(2005) Measuring atrophy in Alzheimer disease: A serial MRI study over 6 and 12 months. Neurology 65, 119–124.

15.

Bard

, Cannon

, Barbour

, Burke

, Games

, Grajeda

, Guido

, Hu

, Huang

, Johnson-Wood

, Khan

, Kholodenko

, Lee

, Lieberburg

, Motter

, Nguyen

, Soriano

, Vasquez

, Weiss

, Welch

, Seubert

, Schenk

, Yednock

(2000) Peripherally administered antibodies against amyloid beta-peptide enter the central nervous system and reduce pathology in a mouse model of Alzheimer disease. Nat Med 6, 916–919.

16.

Schenk

, Hagen

, Seubert

(2004) Current progress in beta-amyloid immunotherapy. Curr Opin Immunol 16, 599–606.

17.

Shankar

, Li

, Mehta

, Garcia-Munoz

, Shepardson

, Smith

, Brett

, Farrell

, Rowan

, Lemere

, Regan

, Walsh

, Sabatini

, Selkoe

(2008) Amyloid-beta protein dimers isolated directly from Alzheimer’s brains impair synaptic plasticity and memory. Nat Med 14, 837–842.

18.

Rinne

, Brooks

, Rossor

, Fox

, Bullock

, Klunk

, Mathis

, Blennow

, Barakos

, Okello

, Rodriguez Martinez de Liano

, Liu

, Koller

, Gregg

, Schenk

, Black

, Grundman

(2010) 11C-PiB PET assessment of change in fibrillar amyloid-beta load in patients with Alzheimer’s disease treated with bapineuzumab: A phase 2, double-blind, placebo-controlled, ascending-dose study. Lancet Neurol 9, 363–372.

19.

Blennow

, Zetterberg

, Rinne

, Salloway

, Wei

, Black

, Grundman

, Liu

(2012) Effect of immunotherapy with bapineuzumab on cerebrospinal fluid biomarker levels in patients with mild to moderate Alzheimer disease. Arch Neurol 69, 1002–1010.

20.

Salloway

, Sperling

, Fox

, Blennow

, Klunk

, Raskind

, Sabbagh

, Honig

, Porsteinsson

, Ferris

, Reichert

, Ketter

, Nejadnik

, Guenzler

, Miloslavsky

, Wang

, Lu

, Lull

, Tudor

, Liu

, Grundman

, Yuen

, Black

, Brashear

(2014) Two phase 3 trials of bapineuzumab in mild-to-moderate Alzheimer’s disease. N Engl J Med 370, 322–333.

21.

McKhann

, Drachman

, Folstein

, Katzman

, Price

, Stadlan

(1984) Clinical diagnosis of Alzheimer’s disease: Report of the NINCDS-ADRDA Work Group under the auspices of Department of Health and Human Services Task Force on Alzheimer’s Disease. Neurology 34, 939–944.

22.

Folstein

, Folstein

, McHugh

(1975) Mini-mental state. A practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res 12, 189–198.

23.

Rosen

, Terry

, Fuld

, Katzman

, Peck

(1980) Pathological verification of ischemic score in differentiation of dementias. Ann Neurol 7, 486–488.

24.

Liu

, Schmidt

, Margolin

, Sperling

, Koeppe

, Mason

, Klunk

, Mathis

, Salloway

, Fox

, Hill

, Les

, Collins

, Gregg

, Di

, Lu

, Tudor

, Wyman

, Booth

, Broome

, Yuen

, Grundman

, Brashear

(2015) Amyloid-β 11C-PiB-PET imaging results from 2 randomized bapineuzumab phase 3 AD trials. Neurology 85, 692–700.

25.

Vanmechelen

, Vanderstichele

, Davidsson

, Van Kerschaver

, Van Der Perre

, Sjögren

, Andreasen

, Blennow

(2000) Quantification of tau phosphorylated at threonine 181 in human cerebrospinal fluid: A sandwich ELISA with a synthetic phosphopeptide for standardization. Neurosci Lett 285, 49–52.

26.

Fox

, Cousens

, Scahill

, Harvey

, Rossor

(2000) Using serial registered brain magnetic resonance imaging to measure disease progression in Alzheimer disease: Power calculations and estimates of sample size to detect treatment effects. Arch Neurol 57, 339–344.

27.

Adedokun

O LM

, Ito

, Raje

, Samtani

, Xu

, Corrigan

, Hu

(2013) Confirmatory population pharmacokinetic analysis of bapineuzumab in subjects with mild to moderate Alzheimer’s disease in two phase 3 studies (ELN115727-301 and ELN115727-302). American Conference on Pharmacometrics, Fort Lauderdale (FL) [poster no W-038].

28.

R Development Core Team (2011) R: A language and environment for statistical computing. R Foundation for Statistical Computing, Austria, Vienna.

29.

Food and Drug Administration. Draft Guidance for Industry: Alzheimer’s Disease: Developing Drugs for the Treatment of Early Stage Disease.: Center for Drug Evaluation and Research, https://www.federalregister.gov/articles/2013/02/08/2013-02863/draft-guidance-for-industry-on-alzheimers-disease-developing-drugs-for-the-treatment-of-early-stage, Last updated February 5, 2013, Accessed on April 10, 2015, Washington, DC.

30.

Salloway

, Sperling

, Gilman

, Fox

, Blennow

, Raskind

, Sabbagh

, Honig

, Doody

, van Dyck

, Mulnard

, Barakos

, Gregg

, Liu

, Lieberburg

, Schenk

, Black

, Grundman

(2009) A phase 2 multiple ascending dose trial of bapineuzumab in mild to moderate Alzheimer disease. Neurology 73, 2061–2070.

31.

Streffer

(2013) Bapineuzumab phase 3 trials in mild-moderate Alzheimer’s disease: Results from the cerebrospinal fluid substudies of 2 phase 3 trials. Alzheimer Association International Conference, Boston, MA.

32.

Jack

Jr , Knopman

, Jagust

, Shaw

, Aisen

, Weiner

, Petersen

, Trojanowski

(2010) Hypothetical model of dynamic biomarkers of the Alzheimer’s pathological cascade. Lancet Neurol 9, 119–128.

33.

Landau

, Lu

, Joshi

, Pontecorvo

, Mintun

, Trojanowski

, Shaw

, Jagust

(2013) Comparing positron emission tomography imaging and cerebrospinal fluid measurements of beta-amyloid. Ann Neurol 74, 826–836.

34.

Storandt

, Head

, Fagan

, Holtzman

, Morris

(2012) Toward a multifactorial model of Alzheimer disease. Neurobiol Aging 33, 2262–2271.

35.

Pimplikar

(2009) Reassessing the amyloid cascade hypothesis of Alzheimer’s disease. Int J Biochem Cell Biol 41, 1261–1268.

36.

Small

, Duff

(2008) Linking Abeta and tau in late-onset Alzheimer’s disease: A dual pathway hypothesis. Neuron 60, 534–542.

37.

Jack

Jr , Knopman

, Jagust

, Petersen

, Weiner

, Aisen

, Shaw

, Vemuri

, Wiste

, Weigand

, Lesnick

, Pankratz

, Donohue

, Trojanowski

(2013) Tracking pathophysiological processes in Alzheimer’s disease: An updated hypothetical model of dynamic biomarkers. Lancet Neurol 12, 207–216.

38.

Fox

, Black

, Gilman

, Rossor

, Griffith

, Jenkins

, Koller

(2005) Effects of Abeta immunization (AN1792) on MRI measures of cerebral volume in Alzheimer disease. Neurology 64, 1563–1572.

39.

Novak

, Einstein

, Tudor

, Gregg

, Collins

, Wyman

, Yuen

, Nielsen

, Grundman

, Brashear

, Fox

, Liu

(2013) The rate of clinical progression and brain atrophy is greater with increasing severity of Alzheimer’s disease: Results from the volumetric MRI substudies of two phase III trials with bapineuzumab. Alzheimers Dement 9(Suppl), P287.