Computer-operated prediction of amyloid-related imaging abnormalities

Abstract

The present application identifies a dose of an active ingredient in therapy provided or considered to be provided to a subject. The level of localized amyloid beta is predicted based on the dose of the active ingredient. The severity of amyloid related imaging anomalies (ARIA) (ARIA-E) exhibiting high signal strength on T2 weighted liquid decay inversion recovery (FLAIR) images is predicted based on the predicted removal of local beta amyloid, wherein predicting the severity of ARIA-E includes predicting the degree of vascular wall disorder. And outputting a result corresponding to the predicted ARIA-E severity degree.

Description

[0001] Cross-references to related patent applications

[0002] This application claims the benefit of and priority to U.S. Provisional Patent Application No. 63 / 337,992, filed on May 3, 2022, which is incorporated herein by reference in its entirety for all purposes. Background Art

[0003] Over the past few decades, amyloid-β (Aβ) aggregates have been the most common target in clinical trials investigating disease-modifying therapies for Alzheimer's disease. Several therapies based on monoclonal antibodies against Aβ aggregates have reported amyloid-associated imaging abnormalities (ARIA). Two types of ARIA have been identified by magnetic resonance imaging (MRI) of the brain. ARIA-E, which appears as high signal intensity on T2-weighted fluid-attenuated inversion recovery (FLAIR) images, indicates vasogenic edema, sulcal fluid accumulation, and gyral swelling. ARIA-H, which appears as low signal intensity on T2*-weighted gradient echo sequences, is thought to represent hemosiderosis and microhemorrhages. ARIA-E is transient and usually resolves within a few months, while ARIA-H remains visible on subsequent MRIs. Most ARIA-E events are mild to moderate in severity and are usually not associated with symptoms, as assessed by various radiological scales that show good correlation.

[0004] The risk of therapy-related ARIA appears to be dose-dependent and is increased in APOEε4 carriers. To reduce the risk of ARIA, several anti-amyloid clinical trials have implemented dose escalation regimens to target doses, as well as routine brain MRI monitoring. The incidence of therapy-related ARIA events in various anti-amyloid clinical trials has been reported to range from 10%-42%, with the highest incidence in APOEε4 allele carriers. Spontaneous ARIA events are rare in placebo groups (e.g., <3% ARIA-E) and outside the context of anti-amyloid clinical trials. An example of the latter is ARIA-like events that occur in cerebral amyloid angiopathy-associated inflammation (CAA-ri), a rare autoimmune response to vascular Aβ aggregates. In clinical practice, ARIA may become a significant burden on patients with Alzheimer's disease and the healthcare system due to MRI monitoring requirements. In addition, ARIA may limit the ability of patients to reach target doses and benefit from disease-modifying therapies that reduce amyloid.

[0005] Therefore, it would be advantageous to be able to generate predictions of therapy-related ARIA events and their severity. BRIEF DESCRIPTION OF THE DRAWINGS

[0006] The present disclosure is described in conjunction with the accompanying drawings:

[0007] Figure 1 The major pathophysiological pathways linking drug-mediated removal of Aβ aggregates to ARIA are delineated.

[0008] Figure 2 An exemplary ARIA-E prediction network including an ARIA-E prediction system and user equipment is depicted.

[0009] Figure 3 The vascular wall disorder (VWD) model and associated illustrative time course and variable relationship graphs are presented.

[0010] Figure 4 Exemplary model performance data at the subject level are presented.

[0011] Figure 5 A flow chart is shown of an exemplary process 400 for predicting ARIA-E according to some embodiments of the invention.

[0012] Figure 6 Scatter plots of exemplary residuals generated by a PK model for predicting drug concentrations in plasma for ARIA-E cases from the SR / MR OLE study are shown. Individual weighted residuals (IWRES) are shown. Solid lines represent empirical (10th, 50th, and 90th) percentiles, while dashed lines represent predicted (10th, 50th, and 90th) percentiles.

[0013] Figure 7 Scatter plots of exemplary residuals generated by the first part (top graph) and the second part (bottom graph) of the two-part PD model for predicting the probability of ARIA-E occurrence and the magnitude of ARIA-E, respectively, are shown. The normalized prediction distribution error (NPDE) and IWRES are also shown. The solid lines represent the empirical (10th, 50th, and 90th) percentiles, while the dashed lines represent the predicted (10th, 50th, and 90th) percentiles.

[0014] FIG. 8A to FIG. 8C Data illustrating the effect of model parameters on amyloid-related imaging abnormalities with edema / effusion (ARIA-E) dynamics are shown. Summary of the invention

[0015] In some embodiments, a computer-implemented method is provided. The method includes a dose of an active ingredient in treatment provided to or considered to be provided to a subject. The method further includes predicting the level of local beta-amyloid based on the dose of the active ingredient, and predicting the severity of amyloid-related imaging abnormalities (ARIA) (ARIA-E) that appear as high signal intensity on T2-weighted fluid-attenuated inversion recovery (FLAIR) images based on the predicted removal of local beta-amyloid, wherein predicting the severity of ARIA-E includes predicting the degree of vascular wall disorder. The method further includes outputting a result corresponding to the predicted severity of ARIA-E.

[0016] The active ingredient may include an anti-amyloid monoclonal antibody.

[0017] The method may include using a pharmacokinetic model to predict a time course of concentration of an active ingredient in the treatment in plasma, wherein the prediction of localized beta amyloid is based on at least one predicted concentration of the active ingredient over the predicted time course.

[0018] Predicting the level of local amyloid beta may include estimating a baseline level of local amyloid beta; calculating a removal rate of local amyloid beta based on a dose of the active ingredient and the baseline level of local amyloid beta; and predicting the level of local amyloid beta based on the calculated removal rate of local amyloid beta.

[0019] Predicting the level of local amyloid beta may include solving a pharmacodynamic differential equation that assumes that the rate of change of amyloid beta is proportional to the product of the concentration of the active ingredient in the subject and the level of local amyloid beta.

[0020] Predicting the level of vessel wall disorder may include solving a differential equation that includes the level of local amyloid beta.

[0021] Predicting the severity of ARIA-E may involve solving an algebraic equation that assumes a nonlinear relationship between the level of vessel wall disorder and the Buchhof Global Score (BGTS).

[0022] The method may include identifying a potential schedule for monitoring for an ARIA event based on a predicted severity of the ARIA, wherein the result characterizes the potential schedule.

[0023] The method may include identifying a potential recommendation for a dosage of an active ingredient for the subject based on the predicted severity of the ARIA, wherein the result characterizes the potential recommended dosage.

[0024] In some embodiments, the system includes a non-transitory computer-readable storage medium containing instructions, which, when executed on the one or more data processors, cause the one or more data processors to perform part or all of one or more methods disclosed herein and / or part or all of one or more processes disclosed herein. Some embodiments of the present disclosure include a computer program product tangibly embodied in a non-transitory machine-readable storage medium, which includes instructions configured to cause one or more data processors to perform part or all of one or more methods disclosed herein and / or part or all of one or more processes disclosed herein.

[0025] The terms and expressions that have been adopted are used as terms of description and not of limitation, and when such terms and expressions are used, there is no intention to exclude any equivalents of the features shown and described or portions thereof. It should be recognized, however, that various modifications are possible within the scope of the claimed system and method. Therefore, it should be understood that although the claimed system and method of the present invention have been specifically disclosed by way of examples and optional features, those skilled in the art should recognize modifications and variations of the concepts disclosed herein, and consider such modifications and variations to be within the scope of the system and method defined by the appended claims.

[0026] The present disclosure is not intended to identify the key or essential features of the claimed subject matter, nor is it intended to be used alone to determine the scope of the claimed subject matter. These illustrative examples are not mentioned to limit or qualify the present disclosure, but to provide examples to aid in understanding the present disclosure. Additional embodiments and examples are discussed in the detailed description, and further description is provided therein. The subject matter should be understood by reference to the appropriate portions of the entire specification of the present disclosure, any or all of the drawings, and each claim.

[0027] The foregoing, together with other features and embodiments, will become more fully apparent with reference to the following description, claims and accompanying drawings. DETAILED DESCRIPTION

[0028] The exact pathophysiology of ARIA remains to be fully elucidated. It has been proposed that drug-mediated removal of Aβ aggregates can increase the permeability of the cerebral vascular wall to fluid and blood products entering the brain (leading to ARIA-E and ARIA-H, respectively). In the brains of patients with Alzheimer's disease, Aβ aggregates are present in the brain parenchyma, as well as in the cerebral vascular walls in the form of cerebral amyloid angiopathy (CAA). Anti-amyloid antibodies associated with ARIA bind to various forms of Aβ aggregates (e.g., oligomers, fibrils, and plaques) and remove them by lysis and / or by phagocytosis mediated by effector cells.

[0029] Figure 1The major pathophysiological pathways thought to link drug-mediated removal of Aβ aggregates to ARIA are delineated. Overall, all postulated pathological mechanisms of ARIA appear to involve some degree of disturbance of the cerebral vascular wall. It is currently unclear to what extent the degree of amyloid burden or the location and rate of amyloid clearance influence the onset, time course, and severity of ARIA. A better understanding of the etiology of ARIA has the potential to reveal additional risk factors, which could accelerate the achievement of patient-centered strategies for dosing and monitoring of individuals with Alzheimer's disease.

[0030] Some embodiments of the present invention relate to predicting the transient dynamics and dose dependence of ARIA-E severity by using a cyborg computer model of ARIA-E, referred to as a vascular wall disorder (VWD) model. For example, the model can receive input data including dose information indicating the amount of anti-amyloid monoclonal antibody (e.g., gantuzumab) administered in a simulation according to a given schedule (e.g., such that the model simulates administration of doses every four weeks), and the model can generate results corresponding to the predicted occurrence and / or severity of ARIA (or ARIA-E) at each schedule in a set of time points relative to a reference point in the schedule (e.g., administration of the first dose). The predicted severity can be used to generate potential monitoring schedules, dosing schedules, and / or dose adjustment criteria. For example, based on an initial dose schedule, a schedule for monitoring and safety data collection can be proposed, and dose adjustment criteria can be defined to recommend a gradual reduction in dose only when an MRI shows at least a defined degree of ARIA-E.

[0031] The framework of the VWD model allows for different biological interpretations of the modeled mechanisms, Figure 1 The VWD model links several pharmacological and biological factors to the observed pharmacokinetics and magnitude of ARIA-E. One biological factor considered in the model is local amyloid, either vascular or parenchymal, whose drug-mediated removal may trigger a cascade of events leading to ARIA-E in one or more regions of the brain. Another biological variable in the model is the level of VWD, a hypothetical measure of susceptibility to fluid leakage into the brain due to disruption of vascular integrity and / or perivascular inflammation caused by amyloid-drug interactions. The VWD model configuration allows for the assessment of potential interactions between the removal of local amyloid and intrinsic vascular repair processes, which may determine or influence the level of VWD, further affecting the magnitude of ARIA-E.

[0032] ARIA-E Prediction Network

[0033] Figure 2An ARIA-E prediction network is depicted including an ARIA-E prediction system 205 and a user device 210. The ARIA-E prediction system 205 controls a VWD model, which includes a pharmacokinetic model component (controlled by a pharmacokinetic model controller 215) and a pharmacodynamic model component (controlled by a pharmacokinetic model controller 220). The VWD model (as in Figure 3 ) shown on the left can be a semi-mechanistic pharmacokinetic-pharmacodynamic model that integrates known and / or inferred processes and / or (e.g., simultaneously) accounts for inter-individual variability in model parameters.

[0034] Each of the pharmacokinetic model controller 215 and the pharmacodynamic model controller 220 can be configured to specify equations used in the model, retrieve stored data (e.g., parameter values) from a memory, and execute corresponding model components. The aggregate model controller 225 can call the pharmacokinetic model controller 215 and / or the pharmacodynamic model controller 220, send input data to it, and / or receive output from it. Therefore, the aggregate model controller 225 can be configured to control the definition and execution of the VWD model by interacting with the pharmacokinetic model controller 215 and / or the pharmacodynamic model controller 220. The aggregate model controller 225 can further perform pre-processing and / or post-processing. For example, such pre-processing and / or post-processing can be adjusted for subject-specific characteristics, such as age, APOE e4 genotype, baseline ARIA-H load, and vascular risk factors (which can be alternatively or additionally factors used by the pharmacokinetic and / or pharmacodynamic models to generate predictions).

[0035] The pharmacokinetic model component is configured to predict the time course of the concentration of the active ingredient in the subject's plasma based on the dosing history and the longitudinal observations of the plasma concentration. An exemplary model may include a pharmacokinetic model disclosed in Retout S et al. "Disease Modeling and Model-Based Meta-Analyses to Define a New Direction for a Phase III Program of Gantenerumab in Alzheimer's Disease." Clin Pharmacol Ther. 2022 April; 111 (4): 857-866, which is incorporated herein by reference in its entirety for all purposes.

[0036] Figure 3The top graph in the middle graph of shows an exemplary time course of the concentration of the active ingredient in the plasma of a subject estimated by using a pharmacokinetic model (left axis) when the active ingredient is administered at a dose indicated by the vertical position of the right axis of the plus sign at a time point indicated by the horizontal position of the plus sign.

[0037] The pharmacodynamic model component is configured to predict ARIA-E observations based on the concentration of the active ingredient based on an intermediate calculation that predicts changes in local amyloid beta levels and vascular wall disturbances triggered by the active ingredient. The intermediate calculation may include solving one or more differential equations and / or performing one or more algebraic calculations. Exemplary equations that may be used in the pharmacodynamic model component are as follows.

[0038] The first exemplary differential equation is based on the assumption that the rate of drug-mediated local Aβ removal depends on the drug concentration in plasma The product of the local Aβ(t) level [arbitrary units] is calculated by the parameter

[0039]

[0040] where t is time, and the initial value of Aβ at the start of OLE treatment (i.e., Aβ(t=0)) is the amyloid 0 For a given model execution, an initial value of Aβ (which may be parenchymal or vascular Aβ) may be estimated or provided depending on the availability of data.

[0041] It is important to note that this equation refers to modeling local levels of amyloid, which represents a hypothetical measure of vascular or parenchymal amyloid from the brain region affected by ARIA-E. The rate of removal of this "local amyloid" is assumed to be proportional to both the drug concentration in plasma and the existing local amyloid level. The proportionality factor, denoted here as α 去除 The removal constant can be interpreted as the rate constant of the biochemical reaction between anti-amyloid antibodies and amyloid aggregates. Such reactions occur in correspondence with drug-mediated amyloid removal and their magnitude may vary from individual to individual. The baseline level of local amyloid (here denoted as amyloid 0 ) influences the initial rate of amyloid removal and may also vary from individual to individual. This model does not include other processes that may alter local amyloid levels, such as physiological amyloid production and endogenous amyloid clearance.

[0042] Figure 3The second figure from top to bottom in the middle figure in shows the estimated time course of local Aβ as estimated by using equation (1). In this particular case, it is estimated that local amyloid decreases over time.

[0043] The second exemplary differential equation assumes that local Aβ removal drives the establishment of VWD(t) [arbitrary units], which is determined by the repair rate k 修复 [sky -1 ] characterized by the first-order vascular repair process offset:

[0044]

[0045] The initial value of VWD at the beginning of OLE treatment (ie, VWD (t=0)) was zero.

[0046] like Figure 7 As shown, the parameter amyloid 0 , α 去除 and k 修复 vary between subjects, and their statistical distribution (typical value and width) is estimated. One, several or all of the remaining model parameters may be constant across individuals.

[0047] Figure 3 The third graph from top to bottom in the middle graph of shows the estimated time course of VWD as estimated by using equation (2). In this particular case, it is estimated that VWD first increases, and then generally decreases.

[0048] The variable "VWD" corresponds to a hypothetical concept that describes the susceptibility of fluid leakage into the brain due to treatment-induced defects in brain vascular and perivascular structures, such as a compromised blood-brain barrier and overburdened drainage pathways around intramural arteries. In the absence of any disturbance, VWD is equal to zero. The development of VWD is driven by drug-mediated amyloid removal and offset by putative intrinsic vascular repair processes. With respect to equation (2), vascular repair is assumed to be a first-order process with a rate constant (k 修复 ) and a half-life that is independent of VWD magnitude. VWD magnitude can be related to a predicted ARIA-E score (BGTS) according to, for example, a sigmoid function. This relationship can ensure that the BGTS is zero or small at low levels of VWD, that the ARIA-E score rises sharply at moderate levels, and that an upper limit is placed on the ARIA-E magnitude (BGTS=60) at high levels of VWD.

[0049] Two additional exemplary algebraic equations relate VWD to the observed ARIA-E magnitude. Using two equations to make this connection can address the complex problem of having a large number of zero BGTS values ​​in the ARIA-E score dataset, where the first equation predicts the probability of an ARIA-E event occurring (i.e., non-zero BGTS) for each BGTS observation, and the second equation quantifies the magnitude of ARIA-E once an event is predicted to have a high probability of occurring.

[0050] The first exemplary algebraic equation deals with the probability p(t) that BGTS is positive (non-zero) under the assumption that the log odds ratio is linearly related to VWD(t) and is:

[0051]

[0052] where β 1 represents the linear factor, and the parameter VWD 50 represents the value of VWD(t) that generates a 50% probability of a positive BGTS. The second exemplary algebraic equation quantifies the BGTS magnitude by using a sigmoid response function

[0053]

[0054] Among them BGTS 最大 represents the maximum ARIA-E score, and EG50 [arbitrary units] represents the value of VWD(t) that causes half-maximal BGTS. In addition, pow is a slope factor that gives the sensitivity of the response to the VWD range by determining the steepness of the VWD-BGTS curve. It is worth noting that different types of BGTS data are used in equations (3) and (4). Figure 3 The S-shaped graph on the right side of shows an exemplary relationship corresponding to equation (4). In equation (3), all BGTS observations are converted to Boolean data indicating whether the observed BGTS is zero or positive, while for equation (4), the BGTS observations are limited to strictly positive values.

[0055] Figure 3 The bottom graph in the middle graph of shows the estimated time course of BGTS for an exemplary example related to the other graphs in the middle graph. The estimated time course of BGTS is determined using equations (3) and (4). In this exemplary example, BGTS increases shortly after the start of the treatment regimen and then generally decreases.

[0056] The aggregate model controller 225 can use the predicted BGTS to facilitate determining a recommended dosing schedule and / or a recommended monitoring schedule for an individual experiencing ARIA-E. For example, the aggregate model controller can coordinate the generation of a time course of predicted BGTS for various prescribed schedules for dosing, various prescribed schedules for monitoring, and / or for various different BGTS cutoffs (thresholds of ARIA-E severity that will result in treatment interruption in the model). Thus, for each prescribed dosing schedule, monitoring schedule, and / or BGTS cutoff, a predicted ARIA-E severity progression (time course of BGTS) can be generated, which can then be output (e.g., via the interface controller 235) to help a user (e.g., a clinician) decide which dosing schedule, monitoring schedule, and / or BGTS cutoff to use in a clinical study and / or real-world use case.

[0057] Thus, the VWD model can interrelate the following factors: (i) gambutaline dose, (ii) drug concentration in plasma, (iii) drug-induced local amyloid removal, (iv) VWD presumed to result in fluid leakage into the brain, (v) intrinsic vascular repair processes, and (vi) BGTS quantifying the severity of ARIA-E. Individual-specific information about a given dose and timing of administration can be provided as input, while parameters related to some or all of the other factors can be estimated.

[0058] The parameter collector 230 initially identifies the parameters of each of the pharmacokinetic model component and the pharmacodynamic model component. In some cases, the parameter collector 230 identifies the value of each of the one or more parameters by querying the external source of the parameter. In some cases, the parameter controller 230 learns the value of each of the one or more parameters by fitting the model component or the VWD model using one or more training data sets. For example, the training data set may include the results from a study in which a group of subjects (e.g., each of the subjects has been diagnosed with Alzheimer's disease) receive a given active ingredient (or an active ingredient corresponding to a specific category of active ingredients) at one or more specific doses and according to one or more dosage schedules, such as gantinostat. The dosage schedule may indicate the relative time of various doses of the active ingredient relative to (e.g.) the initial dose, the most recent dose, or another baseline. In some cases, the dose of the active ingredient administered at each identified administration time in the schedule is the same relative to each other administration time. In some cases, the dose of the active ingredient administered at each of at least two administration times is different. For example, the dose may be gradually increased at at least two doses. The study may have been conducted so that MRIs are performed and assessed periodically to monitor ARIA-E, and so that samples (e.g., blood samples) are collected and assessed periodically to determine the concentration of the active ingredient. The assessment may include assigning a severity to any detected ARIA-E using a scoring technique such as the Barkhof Grand Total Score.

[0059] The study data from the data source can be further filtered to selectively include data corresponding to a subset of subjects, such that the training data set includes the selective data. The subset of subjects can, for example, include subjects in whom ARIA-E development (e.g., corresponding to a severity greater than a given threshold, such as a severity greater than zero) was observed during the study or within one year after the first dose in the study.

[0060] In some cases, parameter collector 230 identifies parameters for a single model component in isolation from other model components.For example, parameters for a pharmacokinetic model component may be defined by performing a Bayesian analysis on the study data.

[0061] In some cases, parameter collector 230 fixes some parameters while fitting other parameters. For example, parameter collector 230 can fix the parameters of the pharmacodynamic model components while fitting the pharmacodynamic model components. Parameter collector 230 can use a nonlinear mixed effects model method to estimate parameter values.

[0062] The parameter collector 230 may use assumptions about the distribution of the variables (e.g., the shape of the distribution), the maximum value, whether the variables are dependent on each other, etc. For example, the parameter controller 225 may define the parameter amyloid 0 、a 去除 and k 修复 Assume lognormal inter-individual variability, estimate its typical value (fixed effect), and estimate the distribution width (random effect). As another example, parameter collector 230 can define parameter β 1 、VWD 50 and pow without any inter-individual variability (i.e. only as fixed effects). It can be shown that VWD(t) is related to amyloid 0 Proportional, meaning the parameter amyloid 0 and EG 50 In fact, it only appears in the expression of BGTS(t) in the form of ratio. To ensure structural identifiability, the parameter collector 230 can set EG 50 =1 without loss of generality. The parameter collector 230 may fix the maximum value of BGTS to 60, consistent with the limitation of the 60-point BGTS severity scale. The residual (unexplained) variability describing the difference between the observed BGTS and the predicted BGTS may be assumed to be independent of the prediction and normally distributed.

[0063] The ARIA-E prediction system 205 includes an interface controller 235 that facilitates communication between the ARIA-E prediction system 205 and the user device 210. The interface controller 235 can be configured to process incoming communications from the user device 210 to, for example, detect a request for a prediction generated by the aggregated model controller (e.g., based on output from the pharmacokinetic model controller 215 and / or the pharmacodynamic model controller 220) and / or detect information to be used as input to a prediction (e.g., identifying a dose and / or dose schedule for treating or being considered for treating a given subject).

[0064] The interface controller 235 may further or alternatively be configured to generate code that can be read by a software application and that can trigger the generation of a presentation or interface (e.g., a web page) that may include or represent one or more predictions generated by the ARIA-E prediction system 205. For example, the interface controller 235 may generate web page code that, when executed by a browser, is configured to present a web page (on a website) that identifies one or more predictions. The website and / or web page may also have initially been operated to collect input for executing the model at the user device 210.

[0065] When user device 210 receives information and / or code from ARIA-E prediction system 205, user device 210 may generate and present a presentation that includes the information and / or data specified by the code to be presented. Thus, user device 210 may present data corresponding to a prediction regarding whether or to what extent a given treatment regimen may result in ARIA-E.

[0066] ARIA-E Prediction Process

[0067] Figure 4 A flow chart of an exemplary process 400 for predicting ARIA-E according to some embodiments of the present invention is shown. At block 402, the aggregate model controller 225 identifies an actual or potential dose of an active ingredient. The identification may include identifying an actual or potential dose that is administered or to be administered at a given time within a treatment regimen (e.g., time zero, 8 weeks after the start of the regimen, once a stable dose is reached, etc.). The dose may be, for example, a dose indicated in a communication received from the user device 210 (e.g., as identified by the interface controller 235), a default dose (e.g., as identified and / or retrieved by the aggregate model controller 225), one or more potential doses (e.g., as identified and / or retrieved by the aggregate model controller 225), and the like.

[0068] At box 404, the pharmacodynamic model controller 220 predicts the level of local amyloid. More specifically, the pharmacodynamic model controller 125 can predict the time course of the concentration of the active ingredient based on the dose of the active ingredient, and the pharmacodynamic model controller 220 predicts the time course of the local amyloid level based on the predicted concentration of the active ingredient. The predicted level of local amyloid can be predicted at the same time as the dose is administered, predicted on a date relative to a specific time when the first (or given) dose is administered, predicted at each of a plurality of time points relative to the administration of the first (or given) dose, etc.

[0069] At block 406, the pharmacodynamic model controller 220 predicts the level of vascular wall disorder based on the predicted local amyloid level and the predicted vascular repair. The predicted level of vascular wall disorder can be generated by solving a differential equation, such as equation (2). The predicted level of vascular wall disorder can further depend on, for example, the predicted repair rate.

[0070] At box 408, the pharmacodynamic model controller 220 predicts ARIA-E severity (e.g., a time course of ARIA-E severity) one or more times. Box 408 may include predicting the level of local amyloid beta at one or more time points (e.g., a time course of levels) and / or using the predicted levels. Box 406 may include predicting the level of vascular wall disorder at one or more time points (e.g., a time course of vascular wall disorder) and / or using the predicted level of vascular wall disorder. Local amyloid beta levels may be predicted using one or more differential equations and based on the estimated active ingredient concentration level. The degree of vascular wall disorder may be predicted using one or more differential equations and based on the predicted local amyloid beta levels (e.g., and / or using the method indicated by reference box 406).

[0071] At block 410, results corresponding to the predicted ARIA-E severity level may be output. For example, the ARIA-E prediction system may transmit the results to the user device 210. The results may include one or more predicted ARIA levels (e.g., a time course of ARIA-E levels), a potential schedule for monitoring MRI-based ARIA-E monitoring (e.g., via a BGTS score), or a potential revised dosing schedule (e.g., indicating one or more recommended doses of an active ingredient and / or a relative timing between one or more consecutive dose administrations).

[0072] Examples

[0073] Dynamic behavior of illustrative cases

[0074] The data used for model development consisted of longitudinal measurements of drug concentrations in plasma and ARIA-E magnitude from a subset of individuals (N=112) who developed ARIA-E during the open-label extension (OLE) of the Scarlet Road (SR; NCT01224106) and Marguerite Road (MR; NCT02051608) studies of gantocorumab, including participants in the prodromal Alzheimer's disease and mild Alzheimer's disease studies, respectively. Briefly, participants in the SR / MR studies who received double-blind treatment and had at least one follow-up visit were eligible to participate in the open-label extension (OLE). During the SR / MR OLE study, all participants received subcutaneous injections of gantocorumab every four weeks and were gradually titrated up to a target dose of 1200 mg. Based on each participant's APOE4 carrier status and the last dose during double-blind treatment, they were assigned to regimen 1 of 5 dose titration regimens (Table 1). Protocol-defined routine MRI monitoring for ARIA-E detection was performed at regular intervals. If ARIA-E is detected, dosing is adjusted based on its radiological severity and / or the presence of symptoms. The magnitude of ARIA-E on MRI scans is rated by the Barkhoff Global Score (BGTS), a 60-point severity scale in which higher scores indicate greater severity. The reported BGTS accounts for the number and size of parenchymal hyperintensities, sulcal hyperintensities, and gyral swellings present on FLAIR images.

[0075]

[0076] choose Figure 3 and Figure 7 Four ARIA-E cases are presented in order to highlight various ARIA-E features (magnitude, resolution, recurrence) that are mechanistically insightful but do not reflect the typical ARIA-E profile of gambuterol. Each study participant was assigned to 1 of the 5 dose titration regimens described in Table 1. Because the model accounts for inter-individual variability, the modeled time distribution for each ARIA-E case was generated based on empirical Bayesian estimates (i.e., the most likely values ​​of individual parameters).

[0077] VWD parameters were estimated by fitting a full VWD model to the pooled set of longitudinal BGTS observations while keeping individual pharmacokinetic parameters fixed. Boolean data from the first part (e.g., “yes / no” of ARIA-E occurrence) and continuous observations from the second part (e.g., BGTS values ​​within a range of ARIA-E magnitudes) were implemented as two different observation types in the data file and modeled simultaneously. Parameter estimation was performed using the nonlinear mixed-effects method of Monolix (version 2020R1).

[0078] Figure 5 The data shown in show that this previously constructed population pharmacokinetic model is able to describe the individual PK data from the SR / MROLE study: the residuals are evenly distributed around zero, and most values ​​are within -2 and +2 standard deviations, indicating the absence of major systematic biases.

[0079] Assumption parameter amyloid 0 , α 去除 and k 修复 With lognormal inter-individual variability, their typical values ​​(fixed effects) and distribution widths (random effects) are estimated. Parameter β 1 、VWD 50 and pow are estimated without any inter-individual variability (i.e. only as fixed effects). It can be shown that VWD(t) is related to amyloid 0 Proportional, meaning the parameter amyloid 0 and EG 50 In fact, it only appears in the expression of BGTS(t) in the form of ratio. Therefore, in order to ensure the identifiability of the structure, EG 50 is set to 1 without loss of generality. In addition, BGTS 最大 The value of was fixed at 60 to fit the limitation of the 60-point BGTS severity scale. The residual (unexplained) variability describing the difference between the observed and predicted BGTS was assumed to be independent of the prediction and normally distributed. The estimated population parameters (typical values ​​and inter-individual variability) are shown in Table 2. Figure 6 The diagnostic plots demonstrate the ability of the model to predict the probability of ARIA-E events at the population level as well as the magnitude of ARIA-E, especially at the individual level: the residuals are uniformly concentrated around zero over time, with no major systematic biases, and most values ​​are within −2 and +2 standard deviations.

[0080]

[0081]

[0082] The estimated time course of active ingredient concentration was calculated using the pharmacokinetic model published by Retout S et al., "Disease Modeling and Model-Based Meta-Analyses to Define a New Direction for a Phase III Program of Gantenerumab in Alzheimer's Disease." Clin Pharmacol Ther. 2022 April; 111(4): 857-866. The estimated time course of local amyloid beta was calculated using equation (1). The estimated time course of VWD was calculated using equation (2). The estimated time course of BGTS was calculated using equations (3) and (4).

[0083] Estimated time course of concentrations, local amyloid-β, VWD, and BGTS in case 1 Figure 3 The middle figure shows that the estimated time course of cases 2-4 is Figure 7 The left part is shown.

[0084] Cases 1 and 2 were assigned to the same rapid up-titration regimen (Regimen 4), but nonetheless developed ARIA-E at significantly different times. (See Figure 3 The middle picture and Figure 7 ). Case 1 reached the target dose by week 16 and experienced the first ARIA-E event at week 20, whereas Case 2 developed ARIA-E at week 12 before reaching the target dose. The two cases appear to experience different rates of amyloid removal around week 12 (as evidenced by the different slopes of the local amyloid curves). This is explained in the model by different estimates of the removal constant (see Figure 7 a in the upper right picture 去除 This difference occurred despite similar levels of drug concentrations during the first 12 weeks and comparable estimates of baseline local amyloid levels.

[0085] Case 3 was assigned to the slowest up-titration regimen (Regimen 1) and underwent two treatment interruptions due to ARIA-E. (See Figure 7 The development of ARIA-E during the slow upward titration period can be attributed to the relatively high baseline level of local amyloid estimated for this subject (see Figure 7 Amyloid protein in the distribution map in the middle right 0The model results suggest that ARIA-E reoccurred upon re-dosing and up-titration, driving another uptick in VWD, through a combination of persistently high levels of local amyloid and relatively high drug concentrations.

[0086] Case 4, assigned to the second slowest up-titration regimen (Regimen 2), developed early ARIA-E that was much higher in magnitude than any ARIA-E episode experienced by the other three cases. (See Figure 7 ) The model did not estimate any significantly different levels of baseline local amyloid or drug concentrations that could explain the apparent differences in the observed ARIA-E magnitudes. However, both the removal and repair constants estimated for Case 4 differed significantly from those estimated for the other three cases. Specifically, the maximum value of the removal constant drives local amyloid removal at a high rate that cannot be matched by the slow rate of vascular repair determined by the small repair rate constant (see Figure 7 Medium α 去除 and k 修复 This imbalance leads to high VWD levels and ultimately high BGTS.

[0087] Model predictions for the four ARIA-E cases shared multiple features. The model predicted a sharp decrease in local amyloid levels around the time of observed ARIA-E events, followed by increases in VWD and BGTS magnitudes. This pattern was evident at week 20 for case 1, week 12 for case 2, weeks 28 and 60 for case 3, and weeks 12 and 52 for case 4 ( Figure 3 and Figure 7 Throughout the treatment interruption period, the model predicted a plateau in local amyloid levels and a gradual reduction in VWD. The latter also reflects the repair process and translates into a gradual reduction in ARIA-E magnitude that closely follows the observed ARIA-E resolution, as can be seen during weeks 20 to 36 in case 1, weeks 12 to 24 in case 2, weeks 28 to 44 in case 3, and weeks 12 to 40 in case 4 ( Figure 3 and Figure 7 ).

[0088] Exploring the impact of key model parameters on ARIA-E dynamics

[0089] Considering that both the baseline level of local amyloid and the removal constant affect the rate of local amyloid removal, FIG. 8A to FIG. 8B The following set of simulations explores their roles in ARIA-E dynamics. Figure 3 Those shown in the central plot of , correspond to case 1. The other plots are simulations with modifications of individual variables ( Fig. 8A The removal constant, Figure 8B Baseline local amyloid-β and Figure 8C The repair rate constant).

[0090] Larger removal constants resulted in earlier increases in both VWD and ARIA-E, whereas smaller removal constants delayed the onset of ARIA-E but also slowed the clearance of local amyloid. Data points at Weeks 12 and 24 provide a comparison. In the setting of high baseline levels of local amyloid, the model predicted early vessel wall disturbances and recurrent ARIA-E events due to persistently high levels of local amyloid. Conversely, in the setting of small baseline levels of local amyloid, the model predicted persistent clearance of local amyloid without any significant ARIA-E events. Figure 8C We show the evolution of ARIA-E when putative vascular repair processes are reduced or enhanced through changes in the repair rate constant. If the repair rate constant of the intrinsic vascular repair process is small, local amyloid removal will lead to repeated vessel wall disturbances and ultimately to recurrent ARIA-E. Conversely, in this model, vascular repair processes with larger rate constants will be able to offset the predicted rate of amyloid removal at the end of the upward titration and prevent any significant ARIA-E.

[0091] explain

[0092] This semi-mechanistic computer model resolves the ambiguity and complexity of ARIA-E pathogenesis in the context of longitudinal observations of ARIA-E magnitude in anti-amyloid clinical trials. Earlier ARIA-E models developed using data from bapinezumab and gantuzumab were event risk models that predicted the probability of ARIA-E with various dosing regimens. The mathematical framework used in this example allows for different biological interpretations of the modeled mechanisms based on key factors: local amyloid burden, drug exposure (dose and drug concentration), amyloid removal, and damage and repair of the vascular wall. For Figure 1In the various ARIA-E scenarios depicted in Figure 1, the same mathematical relationships describe drug-mediated local amyloid removal, regardless of whether the parenchymal or vascular Aβ burden is removed via direct lysis or cell-mediated phagocytosis. Although the precise biological details behind ARIA-E remain to be revealed, a common feature of the different mechanistic pathways that can trigger ARIA-E events appears to be damage to the vascular wall. The integrated model linking drug-mediated local amyloid removal to the level of VWD and ultimately to the severity of ARIA-E (measured by BGTS) performed well in capturing inter-subject variability. Parameters were estimated using a nonlinear mixed-effects approach by fitting the model to the BGTS dataset from the SR / MR OLE study of gantocorumab. The resulting VWD model provided a good subject-level description of the ARIA-E reported in the SR / MR OLE study of gantocorumab, thus confirming the plausibility of the implemented mechanism.

[0093] In addition to fitting the individual time courses of ARIA-E events, the VWD model provides a number of insights. According to the model, the number of ARIA-E events is proportional to the number of events that occur due to (i) high exposure (dose and drug concentration), (ii) high local amyloid, and / or (iii) high efficiency of local amyloid removal (here, captured by the removal constant α). 去除 ), high rates of local amyloid removal may contribute to the pathogenesis of ARIA-E. For anti-amyloid antibodies that rely heavily on Fc receptor-mediated phagocytosis of amyloid aggregates by immune cells, such as aducanumab, donetumab, and gantuzumab, higher estimates of the removal constant may reflect a larger, potentially excessive response of immune cells activated by the amyloid-drug complex. Image-based analysis of a small number of ARIA-E cases from the gantuzumab study revealed that areas of FLAIR high signal intensity often showed a prominent reduction in amyloid PET signal, supporting the notion of more efficient cell-mediated phagocytosis of Aβ aggregates at the ARIA-E site. Notably, significant reductions in amyloid were also observed in areas without any FLAIR abnormalities. Based on the VWD model, focal amyloid reductions can occur without ARIA-E as long as the local rate of amyloid removal is balanced by the rate of intrinsic vascular repair processes. Furthermore, according to the model, ARIA-E can completely resolve before VWD levels return to zero (BGTS=0). As Figure 3As shown, this behavior is a consequence of the putative nonlinear VWD-BGTS relationship, which ensures that small disturbances in the vessel wall do not lead to detectable ARIA-E events. Finally, the VWD model suggests that recurrence of ARIA-E may be a combined effect of high drug concentrations and persistently high levels of local amyloid, and that depletion of amyloid over time ultimately reduces the risk of ARIA-E, even at high drug concentrations.

[0094] When comparing model results to observed amyloid reductions in anti-amyloid clinical trials reporting ARIA, it is useful to distinguish between the global and local nature of the modeled / measured amyloid. For example, studies that performed global PET assessments of baseline amyloid in the bapinezumab and gantuzumab studies indicated that there were no significant differences between the ARIA-E and non-ARIA-E groups. In contrast, when the amyloid PET signal was assessed regionally, the baseline amyloid burden in the occipital brain regions was found to be significantly higher in the ARIA-E group than in the non-ARIA-E group. Given the predilection of CAA for the occipital vasculature, the higher baseline levels of local amyloid estimated by the VWD model may be associated with a higher burden of local vascular amyloid.

[0095] Based on PET imaging alone, it is not possible to determine whether the reduction in amyloid PET signal is due to the removal of vascular and / or parenchymal amyloid. The relative contribution of the two forms of Aβ aggregates to the amyloid PET signal has been controversial. Some studies have concluded that amyloid PET can detect vascular Aβ in CAA, whereas other recent studies have reported that the amyloid PET signal is largely driven by parenchymal amyloid plaques and does not appear to be significantly confounded by CAA. The latter finding is limited by the fact that, although all subjects had pathologically confirmed CAA, only half of the subjects met the clinical diagnosis of suspected CAA, with a median number of two cerebral microbleeds. Of note, a decade ago, the Alzheimer's Association Research Roundtable Task Force recommended that individuals with more than four cerebral microbleeds at baseline should not be enrolled in anti-amyloid clinical trials. Therefore, the contribution of vascular Aβ aggregates to the PET signal in clinically more severe cases of CAA, particularly those enrolled in recent anti-amyloid studies, remains to be determined. To further elucidate the relative risk of ARIA-E associated with regional burden of vascular amyloid and its removal, additional analyses integrating longitudinal PET and MRI data from anti-amyloid trials and quantifying the regional distribution of amyloid signal, FLAIR hyperintensity, and CAA imaging markers are necessary. Future biomarkers that can better distinguish vascular from parenchymal amyloid burden could be used to refine VWD models that could then predict the evolution of ARIA-E in patients with varying degrees of AD pathology (based on parenchymal amyloid burden) and CAA (based on vascular amyloid burden).

[0096] The semimechanistic VWD model provides a mathematical and semimechanistic explanation for the temporal patterns observed in the BGTS data from the gambutin study. The VWD model allows for individual-level exploration in silico of biologically plausible variables and processes that influence the development and severity of ARIA-E. Once validated, the VWD model will help generate hypotheses that can be tested in clinical studies of ARIA management. The VWD model can be used to simulate the effects of the rate of local amyloid removal and the rate of vascular repair on the time course of ARIA-E in individuals with a history of ARIA-E. Such model-based predictions can support decisions to continue or reintroduce therapy in order to minimize the risk of ARIA-E progression.

[0097] Although the subject matter has been described in detail with respect to specific embodiments thereof, it should be understood that those skilled in the art can easily derive variations, changes, and equivalents of such embodiments after understanding the foregoing. Therefore, it should be understood that the present disclosure is presented for purposes of illustration and not limitation, and does not exclude the inclusion of such modifications, changes, and / or additions to the subject matter, which are obvious to those of ordinary skill. In fact, the methods and systems described herein may be embodied in a variety of other forms; in addition, various omissions, substitutions, and changes in the form of the methods and systems described herein may be made without departing from the spirit of the present disclosure. The attached claims and their equivalents are intended to cover such forms or modifications that would fall within the scope and spirit of the present disclosure.

[0098] Unless expressly stated otherwise, it should be understood that throughout this specification, discussions using terms such as "processing," "computing," "calculating," "determining," and "identifying" refer to the action or process by which a computing device (such as one or more computers or one or more similar electronic computing devices) manipulates or transforms data represented as physical electronic or magnetic quantities in a memory, register, or other information storage device, transmission device, or display device of a computing platform.

[0099] One or more systems discussed herein are not limited to any particular hardware architecture or configuration. A computing device may include any suitable arrangement of components that provide a result conditioned on one or more inputs. Suitable computing devices include multi-purpose microprocessor-based computing systems that access stored software that programs or configures the computing system from a general-purpose computing device to a dedicated computing device that implements one or more embodiments of the present subject matter. The teachings contained herein may be implemented in software using any suitable programming, scripting, or other type of language or combination of languages ​​for programming or configuring a computing device.

[0100] Embodiments of the methods disclosed herein may be performed in the operation of such a computing device. The order of the blocks presented in the above examples may be changed—for example, the blocks may be reordered, combined, and / or decomposed into sub-blocks. Certain blocks or processes may be performed in parallel.

[0101] Conditional language used herein, such as, in particular, "may," "could," "might," "can," "e.g.," and the like, unless expressly stated otherwise or otherwise understood in context, is generally intended to convey that some instances include and other instances do not include certain features, elements, and / or steps. Thus, such conditional language is generally not intended to imply that one or more instances in any way require features, elements, and / or steps, or that one or more instances necessarily include logic for determining (with or without author input or prompting) whether such features, elements, and / or steps are included in any particular instance or will be performed in any particular instance.

[0102] The terms "include", "comprises", "have", etc. are synonymous and are used inclusively in an open manner and do not exclude additional elements, features, actions, operations, etc. In addition, the term "or" is used in its inclusive meaning (rather than exclusive meaning), so that, for example, when used to connect a list of elements, the term "or" represents one, some or all of the elements in the list. "Applicable to" or "configured to" used herein is intended to be an open and inclusive language that does not exclude that the device is applicable to or configured to perform additional tasks or steps. In addition, the use of "based on" is intended to be open and inclusive, because the process, step, calculation or other action "based on" one or more listed conditions or values ​​can actually be based on additional conditions or values ​​beyond the listed items. Similarly, the use of "based at least in part" is intended to be open and inclusive, because the process, step, calculation or other action "based at least in part" on one or more listed conditions or values ​​can actually be based on additional conditions or values ​​beyond the listed items. The titles, lists and numbers contained herein are only for ease of explanation and are not intended to be limiting.

[0103] The various features and processes described above can be used independently of each other, or can be combined in various ways. All possible combinations and sub-combinations are intended to fall within the scope of the present disclosure. In addition, some methods or process blocks may be omitted in some specific implementations. The methods and processes described herein are also not limited to any particular order, and the boxes or states associated therewith may be performed in other suitable orders. For example, the blocks or states may be performed in a different order than that specifically disclosed, or multiple blocks or states may be combined in a single block or state. Exemplary blocks or states may be performed serially, in parallel, or in some other manner. Blocks or states may be added to or removed from the disclosed examples. Similarly, the exemplary systems and components described herein may be configured differently from the described systems and components. For example, compared to the disclosed examples, elements may be added to, removed from, or rearranged.

Claims

1. A computer-implemented method comprising: identifying the dose of the active ingredient in the treatment to be provided or contemplated for provision to the subject; predicting a local level of amyloid beta based on said dosage of said active ingredient; predicting the severity of amyloid-related imaging abnormalities (ARIA) (ARIA-E) manifested as high signal intensity on T2-weighted fluid-attenuated inversion recovery (FLAIR) images based on predicted removal of localized beta-amyloid, wherein predicting the severity of ARIA-E includes predicting the degree of vascular wall disorder; as well as The results corresponding to the predicted ARIA-E severity are output.

2. The method of claim 1, wherein the active ingredient comprises an anti-amyloid monoclonal antibody.

3. The method of claim 1 , further comprising using a pharmacokinetic model to predict the time course of concentration of an active ingredient in the treatment in plasma, wherein the prediction of the localized beta-amyloid is based on at least one predicted concentration of the active ingredient over the predicted time course.

4. The method of claim 1, wherein predicting the level of localized amyloid beta comprises: Estimate baseline levels of local amyloid-β; calculating the removal rate of the local amyloid beta protein based on the dose of active ingredient and the baseline level of local amyloid beta protein; as well as The level of local amyloid beta is predicted based on the calculated removal rate of local amyloid beta.

5. The method of claim 1, wherein predicting the level of local amyloid beta comprises solving a pharmacodynamic differential equation that assumes that the rate of change of amyloid beta is proportional to the product of the concentration of the active ingredient in the subject and the level of local amyloid beta.

6. The method of claim 1, wherein predicting the level of vascular wall disorder comprises solving a differential equation including the level of localized beta-amyloid.

7. The method of claim 1, wherein predicting the severity of ARIA-E comprises solving an algebraic equation that assumes a non-linear relationship between the level of vessel wall disorder and the Barkhoff Global Score (BGTS).

8. The method according to claim 1, further comprising: A potential schedule for monitoring an ARIA event is identified based on a predicted severity of the ARIA, wherein the result characterizes the potential schedule.

9. The method according to claim 1, further comprising: identifying a potential recommendation for the dosage of the active ingredient for the subject based on the predicted severity of the ARIA, wherein the result characterizes a potential recommended dosage.

10. A system comprising: one or more data processors; as well as A non-transitory computer-readable storage medium comprising instructions which, when executed on the one or more data processors, cause the one or more data processors to perform part or all of one or more methods disclosed herein.

11. A computer program product tangibly embodied in a non-transitory machine-readable storage medium, comprising instructions configured to cause one or more data processors to perform a portion or all of one or more methods disclosed herein.

Images