This disclosure generally relates to methods of diagnosing an amyloid pathology in the central nervous system of a patient using measurements of enrichment kinetics of at least one amyloid-β isoform. In addition, this disclosure relates to methods of developing and using a mathematical model to predict enrichment kinetics of at least one amyloid-β isoform and to diagnose an amyloid pathology in the central nervous system of a patient using the model.
A paper copy of the sequence listing and a computer readable form of the same sequence listing are appended below and herein incorporated by reference. The information recorded in computer readable form is identical to the written sequence listing, according to 37 C.F.R. 1.821(f).
Alzheimer's Disease (AD) is the most common cause of dementia and is an increasing public health problem. AD, like other central nervous system (CNS) degenerative diseases, is characterized by disturbances in protein production, accumulation, and clearance. In AD, dysregulation in the metabolism of the protein, amyloid-beta (Aβ), is indicated by a massive buildup of this protein in the brains of those with the disease. Because of the severity and increasing prevalence of this disease in the population, it is urgent that better treatments be developed.
The pathogenic causes of Alzheimer's disease are not fully understood, partly due to the difficulty in demonstrating the steps that lead to dementia in humans. Although rare, autosomal dominant AD (ADAD) can be predicted with near 100% certainty in individuals with specific mutations in presenilin 1 (PSEN1), presenilin 2 (PSEN2), or the amyloid precursor protein (APP). Recent findings suggest that a series of ADAD pathophysiological changes occur in the brain decades before clinical dementia manifests. However, the mechanisms by which these mutations lead to AD pathophysiology are not well understood.
The amyloid hypothesis predicts that AD is caused by increased production or decreased clearance of Aβ in the brain, resulting in amyloidosis (the deposition of amyloid proteins in an organ or tissue) and AD's pathologic hallmark of amyloid plaques, which are principally composed of Aβ42. An APP mutation which reduces Aβ production is associated with a strong protective effect against AD, while duplication of APP or mutations which are thought to increase Aβ or Aβ42 cause dominantly inherited AD. Aβ is cleaved from the c-terminal fragment of APP (C99) by PSEN1 and PSEN2, the enzymatic components of gamma-secretase. In cell culture and in plasma, PSEN mutations have been associated with increased Aβ42:Aβ40 ratio, which is hypothesized to increase the risk of amyloidosis. However, others have found that neither the Aβ42:Aβ40 ratio nor Aβ42 levels are increased in vitro. Further, findings of paradoxically reduced cerebrospinal fluid (CSF) Aβ42 concentrations in ADAD patients do not appear to directly support the predicted increased Aβ42 production as an etiologic mechanism in dominantly inherited AD.
Sporadic AD may be characterized by decreased Aβ clearance measured by stable isotope labeling kinetics (SILK). Both sporadic AD and ADAD are associated with lower CSF Aβ42 concentrations and Aβ42:Aβ40 ratios. However, PSEN ADAD mutations are hypothesized to cause increased Aβ42 production, although direct evidence for increased production of Aβ42 in humans has not been reported.
A need exists, therefore, for a method for modeling the in vivo kinetics of Aβ. In particular, a method is needed for modeling the in vivo fractional synthesis rate and clearance rate of proteins associated with a neurodegenerative disease, e.g., the metabolism of Aβ in AD. Such a model may serve as a useful tool in research directed to the characterization and treatment of the underlying processes of AD.
The present disclosure generally relates to systems and methods of modeling and calibrating models for the metabolism and trafficking of CNS biomolecules in a patient.
In one aspect, a method of calibrating a compartmental model for the steady-state kinetics of a biomolecule includes obtaining data values for a level of a labeled moiety in a patient as a function of time. A fraction of the biomolecule is the labeled moiety. The method includes modeling a metabolic pathway of the biomolecule with a compartmental model based on the obtained data values for the labeled moiety, plotting a result of the compartmental model, and comparing a plot of the result to another plot of measured data values. If the plot of the model results matches the other plot of measured data values then the model is sufficiently calibrated. Conversely, if the plot of the result does not match the other plot of the measured data values, then at least one rate constant of the compartmental model is modified. The metabolic pathway of the biomolecule is remodeled using the at least one modified rate constant. The remodeled result of the compartmental model is plotted and compared to the other plot of the measured data values. The actions of comparing and modifying at least one rate constant are repeated, as necessary, to produce a plot the matches the plot of measured data values.
In various other aspects, the method may be performed on one or more computing devices. The computing devices may be distributed across a network or stand-alone devices. In one aspect, the computing device may be used to permit a user to modify and compare plots simultaneously, and in near real time.
In a second aspect, a method for detecting amyloid pathology in the central nervous system of a patient provided that includes: i) determining one or more kinetic parameters of Aβ42 and at least one other Aβ peptide, (ii) comparing the Aβ42 kinetic parameter and the same kinetic parameter for a second Aβ measurement, and (iii) determining whether a subject has amyloid pathology based on a difference between the two kinetic parameters. The kinetic parameter may be selected from the group consisting of fractional synthesis rate, peak time, peak enrichment, initial downturn monoexponential slope, terminal monoexponential slope, and a combination thereof. Two or more kinetic parameters may be determined, three or more kinetic parameters may determined, four or more kinetic parameters may be determined, or at least five kinetic parameters may be determined. The kinetic parameter may be fractional synthesis rate and the Aβ42 fractional synthesis rate may be faster than the fractional synthesis rate for the second Aβ measurement, the kinetic parameter may be peak time and the Aβ42 peak time may be earlier than the peak time for the second Aβ measurement, the kinetic parameter may be peak enrichment and the Aβ42 peak enrichment may be lower than the peak enrichment for the second Aβ measurement, the kinetic parameter may be initial downturn monoexponential slope and the initial Aβ42 slope may be faster than the initial slope for the second Aβ measurement, or the kinetic parameter may be terminal monoexponential slope and the terminal Aβ42 slope may be slower than the terminal slope for the second Aβ measurement. The one or more kinetic parameters may be determined by stable isotope labeling kinetics. A labeled amino acid may be administered to the subject hourly for a time period selected from the group consisting of 6 to 12 hours, 6 to 9 hours, and 9 to 12 hours. The amount of labeled peptide and the amount of unlabeled peptide may be detected by a means selected from the group consisting of mass spectrometry, tandem mass spectrometry, and a combination thereof. The one or more kinetic parameters may be determined using a mathematical model for the enrichment kinetics of Aβ. The method may further include calculating the isotopic enrichment of Aβ42 compared to the second Aβ measurement at a single timepoint after administration of the labeled amino acid to the patient. The second Aβ measurement may be selected from the group consisting of an Aβ peptide other than Aβ42 and total Aβ. The Aβ peptide other than Aβ42 may be Aβ38 or Aβ40. The method may further include (i) calculating the ratio between the Aβ42 kinetic parameter and the same kinetic parameter for the second Aβ measurement, and (ii), comparing the ratio calculated in (i) to a threshold value, wherein a value lower than the threshold indicates the patient has amyloid plaques.
In a third aspect, a method to diagnose an amyloid pathology in a patient is provided. The method includes (i) creating a mathematical model for the steady-state kinetics of Aβ including a set of model parameters (ii) calculating ten times kex42 and adding that to the FTR ratio, and (iii) comparing the value from (ii) to a threshold value, wherein a value lower than the threshold value indicates a subject has Alzheimer's Disease. The set of model parameters includes: kex42, a rate constant for an irreversible loss for Aβ42, and a rate constant for an irreversible loss for Aβ40. kex42 describes the rate of entry of Aβ42 into the exchange compartment and the FTR ratio is the ratio of the rate constants for irreversible loss for Aβ42 versus Aβ40. The amyloid pathology may be selected from the group consisting of amyloid plaques, altered Aβ kinetics (such as Aβ amyloidosis), and Alzheimer's Disease.
In a fourth aspect, a method of calibrating a model to estimate a time course of enrichment kinetics of at least one Aβ isoform is provided. The method includes: a) obtaining data values for an amount of a labeled moiety introduced into a patient as a function of time, wherein a fraction of the at least one Aβ isoforms includes the labeled moiety; b) modeling a metabolic pathway of the at least one Aβ isoform with the model based on the obtained data values to calculate a set of model parameters and an estimated time course of enrichment kinetics of the at least one amyloid; and c) comparing the estimated time course of enrichment kinetics of the at least one Aβ isoform to a measured time course of enrichment kinetics of the at least one Aβ isoform obtained from the patient. If the estimated time course of enrichment kinetics matches the measured time course of enrichment kinetics, the model determines that the compartmental model is calibrated. If the estimated time course of enrichment kinetics does not match the measured time course of enrichment kinetics the model may modify at least one of the set of model parameters and remodel metabolic pathway of the Aβ peptide using the modified model parameters to calculate a new estimated time course of enrichment of the at least one amyloid; these steps may be repeated until the compartmental model is calibrated.
In a fifth aspect, an amyloid kinetics modeling system for estimating a time course of enrichment kinetics of at least one Aβ isoform is provided. The system may include: a) at least one processor; and b) a CRM containing an amyloid kinetics application including a plurality of modules executable on the at least one processor. The plurality of modules may include: i) a plasma module to represent infusion of a labeled moiety into the plasma of a patient and to represent transport of the labeled moiety across the blood brain barrier (BBB) of the patient; ii) a brain tissue module to represent incorporation of the labeled moiety into APP and formation of C99; iii) an amyloid kinetics module to represent cleavage of the C99 to form at least one Aβ isoform and subsequent kinetics of the at least one Aβ isoform within the brain of the patient; iv) a CSF module to represent transport of the at least one Aβ isoform into the CSF of the patient; v) a model tuning module to iteratively adjust a set of model parameters defining a dynamic response of the model to an input time history of plasma leucine enrichment into the plasma module in order to optimize a match between predicted enrichment kinetics and measured enrichment kinetics of the at least one Aβ isoform in the patient; and vi) a GUI module to generate one or more forms used to receive inputs to the system and to deliver output from the system. The plasma module includes a plasma amino acid compartment including a plasma concentration of at least one amino acid, wherein the plasma concentration of the at least one amino acid may be determined using an input including a time history of an infusion of a labeled amino acid into a patient. The brain tissue module includes: a) an APP compartment including a total amount of APP; b) an APP incorporation rate including a rate of incorporation of the at least one amino acid from the plasma amino acid compartment into an APP molecule in the APP compartment; c) a C99 compartment including a total amount of C99 c-terminal fragments; d) a C99 formation rate including a rate of formation of the C99 c-terminal fragments in the C99 compartment from the APP molecules; and e) a C99 clearance rate including a rate of disappearance of the C99 c-terminal fragments from the C99 compartment. The amyloid kinetics module includes: a) a soluble Aβ42 isoform compartment including an amount of a soluble Aβ42 isoform; b) an Aβ42 isoform formation rate including a rate of formation of soluble Aβ42 isoform from the C99 c-terminal fragments; c) an Aβ42 isoform clearance rate including a rate of disappearance of Aβ42 isoforms from the soluble Aβ compartment; d) an Aβ42 incorporation rate including a rate of transformation of the soluble Aβ42 isoform to an incorporated Aβ42 isoform; and e) a recycled Aβ42 compartment including a total amount of incorporated Aβ42 isoform. The CSF module includes a) a CSF Aβ42 compartment including a total amount of CSF Aβ42 isoforms; b) a CSF Aβ42 transfer rate including a rate of transfer of soluble Aβ42 isoform from the soluble Aβ42 compartment to the CSF Aβ42 compartment; and c) a CSF Aβ42 clearance rate including a rate of disappearance of CSF Aβ42 from the CSF Aβ42 pool. The amyloid kinetics module may further include: a) a soluble comparison Aβ isoform compartment including an amount of a soluble comparison Aβ isoform; b) a comparison Aβ isoform formation rate including a rate of formation of soluble comparison Aβ isoform from the C99 c-terminal fragments; and c) a comparison Aβ isoform clearance rate including a rate of disappearance of soluble comparison Aβ isoforms from the soluble comparison Aβ isoform compartment. The CSF module may further include: a) a CSF comparison Aβ isoform compartment including a total amount of CSF comparison Aβ isoforms; b) a CSF comparison Aβ isoform transfer rate including a rate of transfer of soluble comparison Aβ isoform from the soluble comparison Aβ isoform compartment to the CSF comparison Aβ isoform compartment; and c) a CSF comparison Aβ isoform clearance rate including a rate of disappearance of CSF comparison Aβ isoform from the CSF comparison Aβ isoform compartment. The comparison Aβ isoform may be chosen from Aβ38 and Aβ40.
In a sixth aspect, a system for estimating the kinetics of amyloid-beta (Aβ) in the CNS of a patient is disclosed that includes: at least one processor; and a CNS Aβ kinetic model application including a plurality of modules executable using the at least one processor. The modules may include: a) a plasma amino acid module to estimate a plasma amino acid compartment including a plasma concentration of at least one amino acid; b) an APP incorporation module to estimate an APP incorporation rate including a rate of incorporation of the at least one amino acid from the plasma amino acid compartment into an APP molecule in an APP compartment; c) an APP module to estimate the APP compartment including a total amount of APP molecules; d) a C99 formation module to estimate a C99 formation rate including a rate of formation of a C99 c-terminal fragment in a C99 compartment from the APP molecules; e) a C99 clearance module to estimate a C99 clearance rate including a rate of disappearance of the C99 c-terminal fragment from the C99 compartment; e) a C99 module to estimate the C99 compartment including a total amount of the C99 c-terminal fragments; f) a free Aβ formation module to estimate at least one free Aβ isoform formation rate, each free Aβ isoform formation rate including a rate of formation of a free Aβ isoform in a free Aβ compartment from the C99 c-terminal fragments; g) a free Aβ clearance module to estimate at least one free Aβ isoform clearance rate, each free Aβ isoform clearance rate including a rate of disappearance of one of the free Aβ isoforms from the free Aβ compartment; h) a free Aβ module to estimate the free Aβ compartment including the total amount of all free Aβ isoforms; i) a free Aβ recycling module to estimate at least one free Aβ incorporation rate, each free Aβ incorporation rate including a rate of transformation of a free Aβ isoform to an incorporated Aβ isoform in a recycled Aβ compartment, and at least one Aβ recycling rate, each Aβ recycling rate including a rate of recycling an incorporated Aβ isoform in the recycled Aβ compartment back into a free Aβ isoform in the free Aβ compartment; j) a CSF Aβ transfer module to estimate at least one CSF Aβ transfer rate, each Aβ transfer rate including a rate of transfer of one free Aβ isoform from the free Aβ compartment to a CSF Aβ compartment; k) a CSF Aβ clearance module to estimate at least one CSF Aβ clearance rate, each CSF Aβ clearance rate including a rate of disappearance of one CSF Aβ isoform from the CSF Aβ compartment; and I) a CSF Aβ module to estimate the CSF Aβ compartment including the total amount of CSF Aβ isoforms. The Aβ isoforms may be chosen from Aβ38, Aβ40, and Aβ42. At least a portion of the plasma amino acid compartment may include a plasma concentration of at least one labeled amino acid. At least a portion of the APP compartment may include an amount of enriched APP molecules incorporating the at least one labeled amino acid. At least a portion of the C99 compartment may further include an amount of enriched C99 c-terminal fragments formed from the amount of enriched APP molecules. At least a portion of the Aβ isoforms may further include an amount of enriched Aβ isoforms formed from the amount of enriched C99 c-terminal fragments. The CSF Aβ transfer module may further estimate at least one CSF Aβ delay, each CSF Aβ delay including a delay in the transfer of one free Aβ isoform from the free Aβ compartment to the CSF Aβ compartment. The at least one CSF Aβ transfer rate may be represented by a fluid flow of ISF within the brain.
In a seventh aspect, a method of using a model of amyloid β (Aβ) isoform enrichment kinetics is provided that includes: a) obtaining from a patient measured Aβ enrichment kinetics data including a time course of concentration of a labeled moiety infused into the patient, a measured time course of Aβ42 enrichment kinetics in the CSF of the patient, and a measured time course of at least one other comparison Aβ isoform enrichment kinetics in the patient; b) inputting the measured Aβ enrichment kinetics data into the model, wherein the model represents enrichment kinetics of Aβ42 and the at least one other comparison Aβ isoform; c) obtaining a set of model parameters from the model; d) calculating a model index including a mathematical combination of at least two model parameters from the model; e) comparing the model index to a pre-selected threshold range; and f) identifying a disease state of the patient if the model index falls outside of the threshold range. The disease state may be identified as Alzheimer's if the model index falls outside of the threshold range. The severity of the disease state may be identified by comparing the model index to a pre-selected correlation of the disease state with the model index. The correlation of the disease state may be a correlation of the model index with PIB imaging values obtained from a population of patients with a range of disease states. The measured Aβ enrichment kinetics data from a patient may be obtained by the SILK method. The labeled moiety may be labeled leucine. The at least one other comparison Aβ isoform may be chosen from Aβ38 and Aβ40. The model parameters may be chosen from: concentration of Aβ isoforms, rates of transfer, rates of irreversible loss, rates of exchange, rates of delay, and combinations thereof. The model index may be calculated using a rate of irreversible loss of Aβ42 and a rate of transfer of Aβ42. The model parameters may be obtained by iteratively varying the model parameters until a best fit of the estimated Aβ enrichment kinetics to the measured Aβ enrichment kinetics is obtained.
In an eighth aspect, an amyloid kinetics modeling system for estimating a time course of enrichment kinetics of at least one Aβ isoform is provided that includes: a) at least one processor; and b) a CRM containing an amyloid kinetics application including a plurality of modules executable on the at least one processor. The plurality of modules may include: i) a plasma module to represent infusion of a labeled moiety into the plasma of a patient and to represent transport of the labeled moiety across the blood brain barrier (BBB) of the patient; ii) a brain tissue module to represent incorporation of the labeled moiety into APP and formation of C99; iii) an amyloid kinetics module to represent cleavage of the C99 to form at least one Aβ isoform and subsequent kinetics of the at least one Aβ isoform within the brain of the patient; iv) a CSF module to represent transport of the at least one Aβ isoform into the CSF of the patient; v) a blood enrichment module to represent transport of the at least one Aβ isoform into the blood of the patient; v) a model tuning module to iteratively adjust a set of model parameters defining a dynamic response of the model to an input time history of plasma leucine enrichment into the plasma module in order to optimize a match between predicted enrichment kinetics and measured enrichment kinetics of the at least one Aβ isoform in the patient; and vi) a GUI module to generate one or more forms used to receive inputs to the system and to deliver output from the system. The plasma module includes a plasma amino acid compartment including a plasma concentration of at least one amino acid, wherein the plasma concentration of the at least one amino acid may be determined using an input including a time history of an infusion of a labeled amino acid into a patient. The brain tissue module includes: a) an APP compartment including a total amount of APP; b) an APP incorporation rate including a rate of incorporation of the at least one amino acid from the plasma amino acid compartment into an APP molecule in the APP compartment; c) a C99 compartment including a total amount of C99 c-terminal fragments; d) a C99 formation rate including a rate of formation of the C99 c-terminal fragments in the C99 compartment from the APP molecules; and e) a C99 clearance rate including a rate of disappearance of the C99 c-terminal fragments from the C99 compartment. The amyloid kinetics module includes: a) a soluble Aβ42 isoform compartment including an amount of a soluble Aβ42 isoform; b) an Aβ42 isoform formation rate including a rate of formation of soluble Aβ42 isoform from the C99 c-terminal fragments; c) an Aβ42 isoform clearance rate including a rate of disappearance of Aβ42 isoforms from the soluble Aβ compartment; d) an Aβ42 incorporation rate including a rate of transformation of the soluble Aβ42 isoform to an incorporated Aβ42 isoform; and e) a recycled Aβ42 compartment including a total amount of incorporated Aβ42 isoform. The CSF module includes: a) a CSF Aβ42 compartment including a total amount of CSF Aβ42 isoforms; b) a CSF Aβ42 transfer rate including a rate of transfer of soluble Aβ42 isoform from the soluble Aβ42 compartment to the CSF Aβ42 compartment; and c) a CSF Aβ42 clearance rate including a rate of disappearance of CSF Aβ42 from the CSF Aβ42 pool. The amyloid kinetics module may further include: a) a soluble comparison Aβ isoform compartment including an amount of a soluble comparison Aβ isoform; b) a comparison Aβ isoform formation rate including a rate of formation of soluble comparison Aβ isoform from the C99 c-terminal fragments; and c) a comparison Aβ isoform clearance rate including a rate of disappearance of soluble comparison Aβ isoforms from the soluble comparison Aβ isoform compartment. The CSF module may further include: a) a CSF comparison Aβ isoform compartment including a total amount of CSF comparison Aβ isoforms; b) a CSF comparison Aβ isoform transfer rate including a rate of transfer of soluble comparison Aβ isoform from the soluble comparison Aβ isoform compartment to the CSF comparison Aβ isoform compartment; and c) a CSF comparison Aβ isoform clearance rate including a rate of disappearance of CSF comparison Aβ isoform from the CSF comparison Aβ isoform compartment. The blood enrichment module includes: a) a blood Aβ42 compartment including a total amount of blood Aβ42 isoforms; b) a blood Aβ42 transfer rate including a rate of transfer of soluble Aβ42 isoform from the soluble Aβ42 compartment to the blood Aβ42 compartment; and c) a blood Aβ42 clearance rate including a rate of disappearance of blood Aβ42 from the blood Aβ42 pool. The comparison Aβ isoform may be chosen from Aβ38 and Aβ40.
In a ninth aspect, an amyloid kinetics modeling system for estimating a time course of enrichment kinetics of at least one Aβ isoform is provided that may include: a) at least one processor; and b) a CRM containing an amyloid kinetics application including a plurality of modules executable on the at least one processor. The plurality of modules may include: i) a plasma module to represent infusion of a labeled moiety into the plasma of a patient and to represent transport of the labeled moiety across the blood brain barrier (BBB) of the patient; ii) a brain tissue module to represent incorporation of the labeled moiety into APP and formation of C99; iii) an amyloid kinetics module to represent cleavage of the C99 to form at least one Aβ isoform and subsequent kinetics of the at least one Aβ isoform within the brain of the patient; v) a blood enrichment module to represent transport of the at least one Aβ isoform into the blood of the patient; v) a model tuning module to iteratively adjust a set of model parameters defining a dynamic response of the model to an input time history of plasma leucine enrichment into the plasma module in order to optimize a match between predicted enrichment kinetics and measured enrichment kinetics of the at least one Aβ isoform in the patient; and vi) a GUI module to generate one or more forms used to receive inputs to the system and to deliver output from the system. The plasma module includes a plasma amino acid compartment including a plasma concentration of at least one amino acid, wherein the plasma concentration of the at least one amino acid may be determined using an input including a time history of an infusion of a labeled amino acid into a patient. The brain tissue module includes: a) an APP compartment including a total amount of APP; b) an APP incorporation rate including a rate of incorporation of the at least one amino acid from the plasma amino acid compartment into an APP molecule in the APP compartment; c) a C99 compartment including a total amount of C99 c-terminal fragments; d) a C99 formation rate including a rate of formation of the C99 c-terminal fragments in the C99 compartment from the APP molecules; and e) a C99 clearance rate including a rate of disappearance of the C99 c-terminal fragments from the C99 compartment. The amyloid kinetics module includes: a) a soluble Aβ42 isoform compartment including an amount of a soluble Aβ42 isoform; b) an Aβ42 isoform formation rate including a rate of formation of soluble Aβ42 isoform from the C99 c-terminal fragments; c) an Aβ42 isoform clearance rate including a rate of disappearance of Aβ42 isoforms from the soluble Aβ compartment; d) an Aβ42 incorporation rate including a rate of transformation of the soluble Aβ42 isoform to an incorporated Aβ42 isoform; and e) a recycled Aβ42 compartment including a total amount of incorporated Aβ42 isoform. The amyloid kinetics module may further include: a) a soluble comparison Aβ isoform compartment including an amount of a soluble comparison Aβ isoform; b) a comparison Aβ isoform formation rate including a rate of formation of soluble comparison Aβ isoform from the C99 c-terminal fragments; and c) a comparison Aβ isoform clearance rate including a rate of disappearance of soluble comparison Aβ isoforms from the soluble comparison Aβ isoform compartment. The blood enrichment module includes: a) a blood Aβ42 compartment including a total amount of blood Aβ42 isoforms; b) a blood Aβ42 transfer rate including a rate of transfer of soluble Aβ42 isoform from the soluble Aβ42 compartment to the blood Aβ42 compartment; and c) a blood Aβ42 clearance rate including a rate of disappearance of blood Aβ42 from the blood Aβ42 pool. The comparison Aβ isoform may be chosen from Aβ38 and Aβ40.
The application file contains at least one photograph executed in color. Copies of this patent application publication with color photographs will be provided by the Office upon request and payment of the necessary fee.
Corresponding reference characters and labels indicate corresponding elements among the views of the drawings. The headings used in the figures should not be interpreted to limit the scope of the claims.
Provided herein are methods for modeling the in vivo kinetics and metabolism of a CNS biomolecule, in particular one or more amyloid-β (Aβ) isoforms. As used herein, the term “CNS biomolecule” refers to a biomolecule synthesized in the central nervous system (CNS). A skilled artisan will appreciate that while a biomolecule may be synthesized in the CNS, the biomolecule may be transported to other compartments of the body, such that the biomolecule may be detected in the CNS, peripheral nervous system, or outside the nervous system (e.g. in the blood). The kinetic model may be developed and/or calibrated utilizing measured data from patients including, but not limited to the blood and/or the cerebrospinal fluid (CSF) of the patients. Blood, as defined herein, may refer to whole blood, plasma, serum, and any other blood fraction known in the art. This disclosure further provides methods for developing a model by determining and predicting steady state metabolic kinetic parameters. In addition, this disclosure additionally provides methods for modeling in vivo metabolism of one or more Aβ isoforms to determine concentrations of the Aβ isoforms at various states, fractional turnover rates of the one or more Aβ isoforms, and production rates of the one or more Aβ isoforms. Also provided are methods for using the model to identify a patient's disease state and predict aspects of Aβ isoform enrichment kinetics and/or concentrations within a patient. In particular, this disclosure relates to methods of modeling Aβ turnover kinetics in a kinetic model. In an aspect, the kinetic model may be a steady state compartmental model, a flow model, or any combination thereof without limitation. In an aspect, the kinetic model may be used to model the metabolism of any CNS biomolecule.
The method of developing the model may include, but is not limited to, measuring a concentration of a labeled moiety introduced into a patient over a period of time. The labeled moiety may be incorporated into an Aβ precursor within the patient. The method may further include measuring concentrations in a biological sample of the Aβ isoforms incorporating the labeled moiety in the patient, and incorporating the measured data into known or hypothesized relationships and/or metabolic processes. In an aspect, the model may predict the measured values. The model may be developed by calibrating the predicted values against measured values and adjusting a set of model parameters to provide a best fit of the predicted enrichment kinetics of the one or more Aβ isoforms in the CNS to the measured kinetics from the patient. In an aspect, the model may output model parameters specific for each patient.
The concentrations of the one or more Aβ precursors and/or one or more Aβ isoforms and associated metabolic processes in the brain may be represented within the model. In one aspect, this representation within the model may include a compartment, a rate constant, flow equation, and/or any other mathematical representation known in the art without limitation. In an aspect, the concentration in a compartment may be calculated by multiplying the concentration in the previous compartment by a transfer rate constant between the two compartments minus any irreversible loss. Different aspects of the model may be differentiated by different numbers of compartments or types of compartments, the order of the compartments, the equations governing the trafficking and flow of Aβ isoforms, the Aβ isoform being modeled, or any other aspect for modeling the metabolism of a CNS biomolecule.
In another aspect, the kinetic model may represent the movement of soluble Aβ isoforms within the brain as a flow from the ventricles to the brain surface and into the CSF and/or blood. In an aspect, the movement of an Aβ isoform in the brain interstitial fluid (ISF) may be represented by at least one fluid flow equation. In another aspect, the flow of Aβ isoforms may be represented as a transfer between nodes distributed spatially between the point where the Aβ isoform may enter the ISF and the surface of the brain.
In an aspect, the concentration of a labeled moiety and measured concentrations of labeled Aβ isoforms in the CSF and/or blood may be used to develop a model of the metabolism of the labeled Aβ and to determine the rate constants associated with each compartment or flow equation. In addition, the model may be used to calculate predicted concentrations of the Aβ isoforms in the CSF, in the brain, in the blood, or at any other location in a patient. Non-limiting examples of how the model of in vivo Aβ metabolism may be used include identifying the disease state of a patient, fitting a curve of measured data acquired from a patient, predicting the metabolism, processing, and/or concentration of Aβ and its isoforms in a patient, identifying sensitive pathway components to help design drugs or understand a CNS disease, and investigating changes in the kinetics of the isoforms that may be induced by investigational drugs.
Detailed descriptions of various aspects of the methods of modeling the in vivo metabolism of Aβ are provided herein below.
In various aspects, a method to develop a model to represent the synthesis of one or more Aβ isoforms in the central nervous system in vivo and to predict the turnover and production rates of the one or more Aβ isoforms in one or more patients is provided. Data from patients, including time course amounts of a labeled moiety and the concentration of at least one Aβ isoform, may be used in the development of the model.
In various aspects, the model may be used to predict the turnover and production rates of at least one Aβ isoform in a patient. In an aspect, the model may be used to predict the effects of the dysregulation of Aβ isoform turnover and production rates in a subject with Aβ amyloidosis. The term “Aβ amyloidosis” refers to Aβ deposition in a subject that may result from differential metabolism (e.g. increased production, reduced clearance, or both). Aβ amyloidosis is clinically defined as evidence of Aβ deposition in the brain either by amyloid imaging (e.g. PiB PET) or by decreased cerebrospinal fluid (CSF) Aβ42 or Aβ42/40 ratio. See, for example, Klunk W E et al. Ann Neurol 55(3) 2004, and Fagan A M et al. Ann Neurol 59(3) 2006, each hereby incorporated by reference in its entirety. Subjects with Aβ amyloidosis are also at an increased risk of developing a disease associated with Aβ amyloidosis. Diseases associated with Aβ amyloidosis include, but are not limited to, Alzheimer's Disease (AD), cerebral amyloid angiopathy, Lewy body dementia, and inclusion body myositis. Non-limiting examples of symptoms associated with Aβ amyloidosis may include impaired cognitive function, altered behavior, abnormal language function, emotional dysregulation, seizures, dementia, and impaired nervous system structure or function.
In another aspect, the model may be used to predict the effects of the dysregulation of Aβ isoform turnover and production rates resulting from a degenerative disease in a patient. Any degenerative disease characterized by the dysregulation in the turnover and production rate of any CNS biomolecule including, but not limited to at least one Aβ isoform may be predicted using the model without limitation. By way of non-limiting example, Alzheimer's Disease (AD) is a debilitating disease characterized by accumulation of amyloid plaques in the central nervous system resulting from increased production, decreased clearance, or a combination of increased production and decreased clearance of Aβ protein. While AD is an exemplary disease that may be diagnosed or monitored by various aspects of this disclosure, this disclosure is not limited to AD. It is envisioned that the method may be used in modeling the kinetics, diagnosis, and assessment of treatment efficacy of several neurological and neurodegenerative diseases, disorders, or processes including, but not limited to, AD, Parkinson's Disease, stroke, frontal temporal dementias (FTDs), Huntington's Disease, progressive supranuclear palsy (PSP), corticobasal degeneration (CBD), aging-related disorders and dementias, Multiple Sclerosis, Prion Diseases (e.g. Creutzfeldt-Jakob Disease, bovine spongiform encephalopathy or Mad Cow Disease, and scrapie), Lewy Body Disease, and Amyotrophic Lateral Sclerosis (ALS or Lou Gehrig's Disease). It is also envisioned that the method of modeling in vivo kinetics of a CNS disease may be used to study the normal physiology, metabolism, and function of the CNS.
The in vivo metabolism of at least one Aβ isoform or other CNS biomolecule may be modeled in any human patient without limitation. In one aspect, the human patient may be of an advanced age including, but not limited to, human patients older than about 85. Alternatively, the in vivo metabolism of CNS biomolecules may be modeled in other mammalian patients without limitation. In another aspect, the patient may be a companion animal such as a dog or cat. In another alternative aspect, the patient may be a livestock animal such as a cow, pig, horse, sheep or goat. In yet another alternative embodiment, the patient may be a zoo animal. In another aspect, the patient may be a research animal such as a non-human primate or a rodent.
In various aspects, the architecture of the model may be developed using any known or hypothesized pathways and/or mechanisms of Aβ biometabolism without limitation.
Without being limited to any particular theory, amino acids, including labeled amino acids, may be incorporated into amyloid precursor protein (APP) in neural cells. Amyloid precursor protein (APP) is a transmembrane protein expressed in many cells and may be concentrated in neurons and neuronal synapses. APP may be processed by α-, β-, and/or γ-secretases, creating peptides of varying length including, but not limited to, Aβ. C99 forms the c-terminal fragment of APP and is cleaved by the action of β-secretase. Aβ is a peptide of 36-43 amino acids located within the membrane-spanning domain of APP. Aβ is typically formed by the cleavage of APP by the β- and γ-secretases in succession or by the cleavage of C99 by α-secretase. γ-secretase includes enzymatic components PSEN1 and PSEN2. Varying isoforms of Aβ (e.g. Aβ38, Aβ40, Aβ42) may be produced through further processing and cleavage in the endoplasmic reticulum, the trans-Golgi network, or other areas of post-processing.
Because APP and C99 are cell-associated proteins, these proteins are not considered soluble and are not transported within the brain via flow mechanisms. However, after cleavage by α-secretase, Aβ peptides can flow within the brain's interstitial fluid (ISF). The Aβ peptides may be degraded within the brain, taken up in reversible higher order structures (e.g. micelles), taken up irreversibly into plaques, transported across the blood-brain barrier to the blood stream, and/or transported out of the brain as the ISF merges with the CSF, as illustrated in
To track the formation and kinetics of Aβ in vivo, newly formed APP may be labeled by incorporation of a labeled moiety during protein production. The labeled APP may then be cleaved into labeled Aβ isoforms. In an aspect, the labeled moiety may be an amino acid with a stable isotope of carbon, nitrogen, or any other isotope that may be incorporated into amino acids during protein production. Because leucine is more easily capable of crossing the blood brain barrier compared to other amino acids, leucine may be better-suited for use with CNS biomolecules and Aβ. Referring back to
The method for developing a model may include representing the metabolism of any biomolecule derived from the CNS in vivo including, but not limited to, at least one Aβ isoform. The CNS biomolecule may include, but is not limited to, a protein, a lipid, a nucleic acid, a carbohydrate, or any CNS biomolecule known in the art. Any CNS biomolecule may be represented, so long as the CNS biomolecule may be labeled during in vivo synthesis and a sample may be collected from which their metabolism may be measured. In an aspect, the CNS biomolecule is a protein synthesized in the CNS. Non-limiting examples of suitable proteins to be modeled include: amyloid-β (Aβ), Aβ isoforms and other variants, soluble amyloid precursor protein (APP), apolipoprotein E (isoforms 2, 3, or 4), apolipoprotein J (also called clusterin), Tau (another protein associated with AD), glial fibrillary acidic protein, alpha-2 macroglobulin, synuclein, S100B, Myelin Basic Protein (implicated in multiple sclerosis), prions, interleukins, TDP-43, superoxide dismutase-1, huntingtin, and tumor necrosis factor (TNF). Additional CNS biomolecules that may be targeted include products of, or proteins or peptides that interact with, GABAergic neurons, noradrenergic neurons, histaminergic neurons, seratonergic neurons, dopaminergic neurons, cholinergic neurons, and glutaminergic neurons.
The method may model the metabolism of APP in one aspect. In an additional aspect, the CNS biomolecule whose in vivo metabolism is modeled may be amyloid-beta (Aβ) protein. In another aspect, isoforms of Aβ (e.g., Aβ40, Aβ42, Aβ38 and/or others) may be modeled. In a further aspect, digestion products of Aβ (e.g., Aβ6-16, Aβ17-28) may be modeled. In an aspect, the model may represent the metabolism of more than one CNS biomolecule at a time. In one aspect, the CNS biomolecule may include, but is not limited to, C99, APP, Aβ38, Aβ40, Aβ42, and any other Aβ isoform.
In an aspect, the plasma concentration of a labeled moiety may be input into the model. In one aspect, the labeled moiety plasma concentration may be used to develop the model and determine the model parameters.
When the method is employed to model the metabolism of a protein, the labeled moiety may be an amino acid. Those of skill in the art will appreciate that at least several amino acids may be used to provide the label of a CNS biomolecule. Generally, the choice of amino acid is based on a variety of factors such as: (1) the amino acid generally is present in at least one residue of the protein or peptide of interest; (2) the amino acid is generally able to quickly reach the site of protein synthesis and rapidly equilibrate across the blood-brain barrier; (3) the amino acid ideally may be an essential amino acid (not produced by the body), so that a higher percent of labeling may be achieved; (4) the amino acid label generally does not influence the metabolism of the protein of interest (e.g., very large doses of leucine may affect muscle metabolism); and (5) the relatively wide availability of the desired amino acid (i.e., some amino acids are much more expensive or harder to manufacture than others).
In an aspect, the amino acid leucine may be used to label proteins that are synthesized in the CNS. Non-essential amino acids may also be used; however, measurements may be less accurate. In one aspect, 13C6-phenylalanine, which contains six 13C atoms, may be used to label a neurally derived protein. In an aspect, 13C6-leucine may be used to label a neurally derived protein. In an exemplary aspect, 13C6-leucine may be used to label amyloid-β.
There are numerous commercial sources of labeled amino acids, containing both non-radioactive isotopes and radioactive isotopes. Generally, the labeled amino acids may be produced either biologically or synthetically. Biologically produced amino acids may be obtained from an organism (e.g., kelp/seaweed) grown in an enriched mixture of 13C, 15N, or another isotope that is incorporated into amino acids as the organism produces proteins. The amino acids are then separated and purified. Alternatively, amino acids may be made using any known synthetic chemical processes. The labeled moiety may be administered to a patient using any one of at least several methods known in the art. Non-limiting examples of suitable methods of administration include intravenous, intra-arterial, subcutaneous, intraperitoneal, intramuscular, and oral administration. In one aspect, the labeled moiety is administered to the patient using intravenous infusion.
The labeled moiety may be administered slowly over a period of time or as a large single dose depending upon the type of analysis chosen (e.g., steady state or bolus). To achieve steady-state levels of the labeled CNS biomolecule, the labeling time generally should be of sufficient duration so that the labeled CNS biomolecule may be reliably quantified. The labeling time sufficient for reliable quantification of steady state levels of a labeled Aβ in a blood sample is typically less than required time for reliable quantification of steady state levels of Aβ in a CSF sample. See for example, U.S. Pat. No. 7,892,845 and U.S. Ser. No. 13/669,497, each hereby incorporated by reference in its entirety. This duration may be selected to be sufficient to result in saturation of the biochemical pathways associated with the synthesis of the CNS biomolecule. In one aspect, the duration may be sufficient to result in the saturation of the biochemical pathways associated with the synthesis and kinetics of at least one Aβ isoform in the brain of a patient, including, but not limited to: APP synthesis, cleavage of C99 and the at least one Aβ isoform, the transport of the at least one Aβ isoform to the CSF, and the transport of the at least one Aβ isoform to the blood. In another aspect, the saturation of the biochemical pathways may be indicated by the detection of stabilized levels of the at least one Aβ isoform in the CSF and/or blood as measured in a patient.
In an aspect, the labeled moiety is administered intravenously for an amount of time that is less than the half-life of Aβ in blood or CSF. In other aspect, the labeled moiety is administered intravenously for an amount of time that is greater than the half-life of Aβ in blood or CSF. For example, the labeled moiety may be administered intravenously over a duration of minutes to hours, including, but not limited to, for at least 10 minutes, at least 20 minutes, at least 30 minutes, at least 1.0 hour, at least 1.5 hours, at least 2.0 hours, at least 2.5 hours, at least 3.0 hours, at least 3.5 hours, at least 4.0 hours, at least 4.5 hours, at least 5.0 hours, at least 5.5 hours, at least 6.0 hours, at least 6.5 hours, at least 7.0 hours, at least 7.5 hours, at least 8.0 hours, at least 8.5 hours, at least 9.0 hours, at least 9.5 hours, at least 10.0 hours, at least 10.5 hours, 1 at least 1.0 hours, at least 11.5 hours, or at least 12 hours. In another aspect, the labeled moiety may be administered intravenously over a period ranging from about 6 hours to about 18 hours. In another aspect, the labeled moiety may be administered intravenously over a period of about 9 hours. In another aspect, the labeled moiety may be administered intravenously over a period of about 3 hours. In yet another aspect, a labeled moiety is administered orally as multiple doses. The multiple doses may be administered sequentially or an amount of time may elapse between each dose. The amount of time between each dose may be a few seconds, a few minutes, or a few hours. In each of the above embodiments, the labeled moiety can be labeled leucine, labeled phenylalanine, or any other labeled amino acid that is capable of crossing the blood-brain barrier.
Those of skill in the art will appreciate that the amount (or dose) of the labeled moiety can and will vary. Generally, the amount is dependent on (and estimated by) the following factors. (1) The type of analysis desired. For example, to achieve a steady state of about 15% labeled leucine in plasma requires about 2 mg/kg/hr over 9 hr after an initial bolus of about 3 mg/kg over 10 min. In contrast, if no steady state is required, a bolus of labeled leucine (e.g., about 400 mg to about 800 mg of labeled leucine) may be given. (2) The Aβ variant under analysis. For example, if the Aβ variant is being produced rapidly, then less labeling time may be needed and less label may be needed—perhaps as little as 100 mg or less as a bolus. And (3) the sensitivity of the technology to detect label. For example, as the sensitivity of label detection increases, the amount of label that is needed may decrease.
In another aspect, a labeled moiety is administered orally as a single bolus. In another aspect, a labeled moiety is administered intravenously as a single bolus. In still another aspect, a labeled moiety is administered intravenously as an infusion for about 1 hour. All three methods of administration (oral bolus, IV bolus, and IV infusion) work equally well in terms of providing a reliable measure of amyloid beta metabolism. An intravenous bolus of a labeled moiety and an oral bolus of labeled moiety are easier to administer than an intravenous infusion, and also results in maximal levels of free label at an earlier time point (e.g. about 5 to about 10 minutes, and about 30 to about 60 minutes, respectively, for labeled leucine). In each of the above embodiments, the labeled moiety can be labeled leucine, labeled phenylalanine, or any other labeled amino acid that is capable of crossing the blood brain barrier.
The method of developing the model may include obtaining a biological sample from a patient so that the in vivo metabolism of the labeled CNS biomolecule may be determined. Information from a patient's biological sample may be used as an input in the method of developing and/or calibrating a model of in vivo metabolism of a CNS biomolecule.
Suitable biological samples include, but are not limited to, cerebral spinal fluid (CSF), blood plasma, blood serum, urine, saliva, perspiration, and tears. In one aspect, biological samples may be taken from the CSF. In an alternate aspect, biological samples may be collected from the urine. In another aspect, biological samples may be collected from the blood.
Cerebrospinal fluid may be obtained by lumbar puncture with or without an indwelling CSF catheter (a catheter is preferred if multiple collections are made over time). Blood may be collected by veni-puncture with or without an intravenous catheter. Urine may be collected by simple urine collection or more accurately with a catheter. Saliva and tears may be collected by direct collection using standard good manufacturing practice (GMP) methods.
In general, when the CNS biomolecule is a protein, the method of developing and/or calibrating the model may include obtaining a first biological sample to be taken from the patient prior to administration of the labeled moiety to provide a baseline for the patient. After administration of the labeled amino acid or protein, one or more samples generally may be taken from the patient. As will be appreciated by those of skill in the art, the number of samples and when they may be taken generally depend upon a number of factors such as: the type of analysis, type of administration, the protein of interest, the rate of metabolism, the type of detection, etc.
In general, samples obtained during the labeling phase may be used to determine the rate of synthesis of the Aβ variant, and samples taken during the clearance phase may be used to determine the clearance rate of the Aβ variant. Labeled Aβ increases during labeling and then decreases after the labeling has stopped. In one aspect, the CNS biomolecule may be a protein including, but not limited to at least one Aβ isoform and one or more samples of CSF may be taken hourly for 36 hours. Alternatively, the samples may be taken every other hour or even less frequently. Typically, biological samples obtained during the first 12 hours of sampling (i.e., 12 hrs after the start of labeling (IV bolus or infusion) may be used to determine the rate of synthesis of the protein, and biological samples taken during the final 12 hours of sampling (i.e., 24-36 hrs after the initial infusion of labeled moieties) may be used to determine the clearance rate of the protein. In another aspect, a single sample may be taken after labeling for a period of time, such as 12 hours, to estimate the synthesis rate, but this may be less accurate than multiple samples. In another aspect, the CNS biomolecule may be a protein including, but not limited to at least one Aβ isoform and one or more samples of blood may be taken hourly for 24 hours. Alternatively, the samples may be taken every other hour or even less frequently. Typically, blood samples obtained during the first 4 hours of sampling (i.e., about 1 minute to about 4 hrs after administration of an IV or oral bolus, about 10 minutes to about 4 hrs after administration of an IV or oral bolus, about 30 minutes to about 4 hrs after administration of an IV or oral bolus, about 1 minute to about 3 hrs after administration of an IV or oral bolus, about 10 minutes to about 3 hrs after administration of an IV or oral bolus, or about 30 minutes to about 3 hrs after administration of an IV or oral bolus) may be used to determine the rate of synthesis of the protein, and blood samples taken during the final 20 hours after administration of an IV or oral bolus (i.e., about 4 hours to about 12 hours after administration of an IV or oral bolus, about 12 hours to about 24 hours after administration of an IV or oral bolus, about 18 hours to about 24 hours after administration of an IV or oral bolus, or about 4 hours to about 24 hours after administration of an IV or oral bolus) may be used to determine the clearance rate of the protein. In another aspect, a single sample may be taken after administration of an IV or oral bolus, such as at about 3 hours, to estimate the synthesis rate, but this may be less accurate than multiple samples. In yet a further aspect, samples may be taken from an hour to days or even weeks apart depending upon the protein's synthesis and clearance rate.
The method of developing a kinetic model may include developing a model that may fit experimental findings in a manner consistent with known molecular biology and physiologic structures. In an aspect, the kinetic model may be a comprehensive steady state compartmental model that uses tracer kinetics to determine the rate constants within the model. In another aspect, the model may account for the time course of at least one Aβ isoform in vivo. In yet another aspect, the model may mathematically represent the one-dimensional flow of soluble Aβ isoforms in the brain from the ventricles to the CSF and/or blood. In this aspect, the flow may be due to the pressure difference between the ventricles and the brain surface.
The plasma module 1608 represents the infusion of the labeled moiety into the plasma of a patient and the transport of the labeled moiety across the blood brain barrier (BBB). The brain tissue module 1610 represents the incorporation of the labeled moiety into APP and the formation of C99. The amyloid kinetics module 1612 represents the cleavage of the C99 to form at least one Aβ isoform and the subsequent kinetics of the at least one Aβ isoform within the brain including, but not limited to, recycling, fractional turnover, incorporation into plaques, transport across the blood brain barrier (BBB), and breakdown of the at least one Aβ isoform. The CSF module 1614 represents transport of the at least one Aβ isoform into the CSF. The model tuning module 1616 may iteratively adjust a set of parameters defining the dynamic response of the model to the input time history of plasma leucine enrichment into the plasma module 1608 in order to optimize the match between the predicted CSF enrichment kinetics and the measured CSF enrichment kinetics of the at least one Aβ isoform in the patient.
In an aspect, the amyloid kinetics application 1606 may further include a blood enrichment module (not shown). The blood enrichment module represents transport of the at least one Aβ isoform into the blood. In an additional aspect, the amyloid kinetics application 1606 may include the blood enrichment module in the place of the CSF module 1614.
The GUI module 1618 may generate one or more forms to receive inputs to the system 1600 such as the time history of plasma leucine enrichment and the measured CSF enrichment kinetics of the at least one Aβ isoform in the patient. The GUI module 1618 may further receive additional user inputs such as defined ranges for parameters defining the dynamic response of the model and other values used to specify the operation of the system 1600. The GUI module 1618 may also generate one or more forms used to display outputs of the application 1606 including, but not limited to graphs of the predicted CSF enrichment kinetics of the at least one Aβ isoform, listings of model parameters, predictions of a disease state of a patient, and any other relevant output.
Any method of modeling may be used to implement any one or more of the modules 1608-1614 without limitation. Non-limiting examples of suitable modeling methods include compartmental models, flow models, mathematical equations, fluid dynamic flow equations, diffusion equations, any other suitable modeling method known in the art. In one aspect, the modules 1608-1614 may be implemented using compartmental models. In another aspect, the modules 1608-1614 may be implemented using a combination of compartmental models and flow models.
(i) Compartmental Model
The kinetic model may account for the full time course of Aβ38, Aβ40, and Aβ42 enrichments and CSF concentrations in one aspect. In an aspect, the model may describe fundamental processes that affect Aβ kinetics including, but not limited to: production, reversible exchange, and irreversible loss, and may account for the effect of the kinetics of these processes on CSF concentrations of Aβ.
The model may be implemented on any software or device without limitation. In an aspect, modeling may be performed using SAAM II software (Resource for Kinetic Analysis, University of Washington, Seattle). In various aspects, the number, order, and location of compartments may vary. In various other aspects, the interconnections between the various compartments may vary. In various additional aspects, functions other than first-order rate constants may be used to represent the movement of a quantity from one compartment to another. Non-limiting examples of suitable functions include linear functions, exponential functions, differential equations, logarithmic equations, and any other known kinetic and/or rate equation known in the art. The functions may be constant with respect to other variables within the model, or the functions may include other variables generated within the model. For example, the rate of synthesis of an Aβ isoform may be influenced by the concentration of soluble Aβ isoform already produced in an aspect.
The kinetic model may include a compartment for the concentration of a labeled moiety. In one aspect, the kinetic model may include a compartment for labeled plasma leucine. In another aspect, the kinetic model may include at least one compartment for APP. In other aspects, the kinetic model may include compartments for iAPP and mAPP. In yet another aspect, the kinetic model may include a compartment for C99. The kinetic model may include parallel arms for different CNS biomolecules or Aβ isoforms. In an aspect, the kinetic model may include three parallel arms with corresponding compartments, one for each Aβ isoform (Aβ42, Aβ40, Aβ38), as illustrated in
The method of developing the kinetic model may include acquiring data from various patients to input into the development of the model. In one aspect, the enrichment of the labeled moiety and labeled Aβ isoform peptides may be measured at frequent time intervals (indicated by solid triangles in
Referring to
The kinetic model may take into consideration that some of the C99 and soluble Aβ peptides may be metabolized to fates other than Aβ peptides that appear at the CSF sampling site in an aspect. Without being limited to any particular theory, the physiologic nature of these other losses for soluble Aβ peptides may be unknown at this time, but the model may include all processes that remove soluble peptides irreversibly, e.g. deposition into plaques, cellular uptake, proteolytic degradation, and/or transfer into the blood. In an aspect, the model may include an irreversible loss of each soluble Aβ isoform that was not recovered in CSF.
In an aspect, a reversible exchange compartment in exchange with the “soluble” Aβ peptide may be added to the model to optimally fit the sigmoid shape of the CSF Aβ enrichment time courses after the peak enrichment. The reversible exchange may represent possible recycling of Aβ isoforms to and/or from plaques, the exchange of labeled Aβ for unlabeled Aβ, the recycling of higher order Aβ structures, or any other reversible exchange of Aβ. In one aspect, a reversible exchange compartment may be included for Aβ42. In an aspect, the exchange process may only be added for an isoform if it improves the Akaike Information Criteria (AIC) of the fit as provided by SAAM II software.
In another aspect, a scaling factor (SF) may be applied to each of the Aβ isoform enrichments after the kinetic model has first been developed if it improves the AIC. Without being limited to any particular theory, the SF may account for small amounts of isotopic dilution between plasma leucine and the biosynthetic precursor pool (generally less than about 5%) or to correct for minor calibration errors (generally less than about 10%) in the measurement of isotope enrichments of plasma leucine and/or Aβ peptides.
One principle parameter obtained with the model is the fractional turnover rate (pools/h) of the “soluble” Aβ peptides, i.e. the sum of the fractional rate of loss of these compartments to CSF and other losses from the system. Based on the calibrated kinetic parameters that describe the shape and magnitude of the CSF Aβ enrichment time course, the model may determine the rate constant (pools/h) for production of each Aβ peptide isoform from their common C99 precursor to accurately project the measured baseline CSF Aβ peptide concentrations. The model may project the steady state masses (ng) within and the flux rates (ng/h) between all compartments for each Aβ isoform.
The rate constants for transfer between compartments in the model may be calibrated for each patient by utilizing the labeled moiety time course and the measured time course of the Aβ isoforms in the biological sample. The model parameters to be calibrated may include, but are not limited to, transfer rate constants for APP, C99, Aβ38, Aβ40, and Aβ42; irreversible loss rate constants for C99, Aβ38, Aβ40, Aβ42, and CSF; exchange rate constants for Aβ38, Aβ40, and Aβ42; return rate constants; delay rate constants; and scaling factors. In another aspect, a database, similar to the data source shown and described below with reference to
Referring to
In an aspect, the model may include a representation of transfers of at least a fraction of the Aβ isoforms in the brain to the CSF. In an aspect, the model may include a representation of transfers of at least a fraction of the Aβ isoforms in the brain to the blood. In an additional aspect, the model may include a representation of transfers of at least a fraction of the Aβ isoforms in the brain to the CSF as well as a representation of transfers of at least an additional fraction of the Aβ isoforms in the brain to the blood.
In various aspects, the architecture of the model may be developed using the data measured from various patients as described above. The results of alternative model architectures that may vary in the number, order, location, and/or interconnections between compartments may be compared using a figure of merit, and the model architecture associated with the most favorable figure of merit may be selected. Non-limiting examples of suitable figures of merit include Akaike information criterion, Bayesian information criterion, Deviance information criterion, Focused information criterion, Hannan-Quinn information criterion, and any other suitable figure of merit known in the art.
Those skilled in the art will recognize that the order of the compartments in a linear model does not affect the fit of the data or the values of the determined parameters. Those skilled in the art will also recognize that some distinct rate constants in these mathematical models may be set to the same value in some cases where the individual parameters are unidentifiable or poorly identified by the data. The impact of these small changes to the structure of the model on the values of the rate constants may typically be minimal.
(ii) Flow Model
In an aspect, the kinetic model may be a flow model.
The kinetic model may account for one-dimensional flow of Aβ isoforms in the ISF of the brain from the ventricles to the brain surface and into the CSF through a pressure differential as illustrated in
In an aspect, Illustrated in
Each node may be divided into an immobile and mobile portion, with the immobile portion remaining at that location in the brain and the mobile portion moving toward the surface of the brain at a velocity that may be derived from the computed velocity summarized in
The irreversible loss of each Aβ isoform may occur simultaneously with the flow of the Aβ isoform in the ISF. The movement of an Aβ isoform at any node or position within the ISF may be described in terms of flow and reaction. The reactions may be defined by the production of the Aβ isoform from C99, the degradation of the Aβ isoform (irreversible loss), and the exchange of the Aβ isoform with immobile forms of the Aβ isoform. In an aspect, each Aβ isoform may be tracked spatially in one dimension and the addition and removal of the Aβ isoform may be accounted for at each x location.
In another aspect, the flow may be incorporated into a compartment or rate constant within the compartmental model. The kinetic model may account for three-dimensional flow of Aβ in one aspect.
In various aspects, the methods for modeling the in vivo metabolism of at least one CNS biomolecule may be performed on one or more processing systems having one or more processors. In an aspect, the CNS molecule may be Aβ or an Aβ isoform. In one aspect, an Aβ modeling calibration system provides one or more graphical user interfaces that enable users to selectively calibrate a modeling system to identify, track, and estimate amounts or levels of a particular Aβ isoform or labeled protein segments at various time points in the metabolic pathway of Aβ. The Aβ modeling calibration system may be used to refine and calibrate a kinetic model for estimating amounts of Aβ lost to: degradation, formation of higher order structures and insoluble plaques, or Aβ otherwise transported to the blood or CSF. The Aβ modeling calibration system may therefore be used to calibrate a model for determining or predicting the fractional turnover rate of the “soluble” Aβ peptides (pools/h). In particular, by comparing model-derived data with known data values stored in memory, in a database, or in any other data storage medium, the system 100 may be used to calibrate the kinetic parameters, also stored in memory, for predicting various rate constants for the metabolism of Aβ peptides based on the measured baseline CSF Aβ peptide concentrations. As previously described, the CNS Aβ modeling calibration system 100 may be used to calibrate the optimal rate constants for the transfer between the various compartments in the kinetic models 10, 20, and 50 by comparing measured labeled moiety concentrations and the measured concentrations of the Aβ isoforms in a biological sample. Moreover, the system 100 may determine or predict the steady state masses (ng) within and the flux rates (ng/h) between the compartments of the model, as shown in
Other aspects of the Aβ modeling calibration system enable users to interact with one or more graphical user interfaces to view and calibrate the optimized rate constant values, predicted fractional turnover rates, or in some embodiments, the kinetic model itself. The Aβ modeling calibration system 100 enables a user to select and manually or automatically adjust or modify one or more input values or rate constant values of the kinetic model.
As shown in
Each processing device 102 or 108 may also include a stand-alone or distributed version of the MCA application 104, to generate one or more graphical user interface(s) 120 on the display 114. The graphical user interface 120 enables a user of the processing devices 102 or 108 to view actual experimental data, predicted data, and other data manually input using the input device 116 or otherwise stored in the data source 106. The graphical user interface 120 also enables a user of the processing devices 102 or 108 to view and modify the stored data as well as any determined or predicted data values. According to another aspect, the graphical user interface 120 enables a user of the MCS system 100 to interact with various data entry forms to enter authentication data or other data, including but not limited to usernames, passwords or other user data, to access any restricted functionality of the MCS 100.
According to one aspect, the computing device 102 includes a computer-readable medium (“CRM”) 122, also referred to herein as a machine-readable medium, configured with the MCA 104. The CRM 122 includes instructions or modules that are executable by the processor(s) 112. The CRM 122 may include volatile media, nonvolatile media, removable media, non-removable media, and/or another available medium that can be accessed by the computing device 122. By way of example and not limitation, the CRM 122 comprises computer storage media and communication media. Computer storage media includes non-transient memory, volatile media, nonvolatile media, removable media, and/or non-removable media implemented in a method or technology for storage of information, such as computer readable instructions, data structures, program modules, or other data. Communication media may embody computer readable instructions, data structures, program modules, or other data and include an information delivery media or system
The data source 106 may be a database or other general repository of data including, but not limited to, MCS user data, patient data, model data, or any other data. The data source 106 or database may include memory and one or more processors or processing systems to receive, process, query, and transmit communications or requests to store and/or retrieve such data. In another aspect, the database may be a database server.
Similarly, the local or client computing device 108 may be a processing device similar to the processing device 102, one or more servers, personal computers, mobile computers, and other computing devices. In various aspects, the local computing devices 108 include one or more processors and volatile and/or non-volatile memory and may be configured to communicate over the communication network 112 via wireless and/or wireline communications.
The computing device 102 may be configured to receive data and/or communications from and/or transmit data and/or communications to a client 108 or other computing device, including a remote data source through the communication network 112. The communication network 112 can be can be the Internet, an intranet, and/or another wired and/or wireless communication network. In one aspect, the computing device 102, the client 108, and/or the data source 106 communicate data in packets, messages, or other communications using a protocol, such as a Hypertext Transfer Protocol (HTTP) or a Wireless Application Protocol (WAP). Other examples of communication protocols exist.
Example embodiments of the methods and systems described herein may be implemented at least in part in electronic circuitry; in computer hardware executing firmware and/or software instructions, such as the MCA 104; and/or in combinations thereof. Example embodiments also may be implemented using a computer program product (e.g., a computer program tangibly or non-transitorily embodied in a machine-readable medium and including instructions for execution by, or to control the operation of, a data processing apparatus, such as, for example, one or more programmable processors or computers). A computer program may be written in any form of programming language, including compiled or interpreted languages, and may be deployed in any form, including as a stand-alone program or as a subroutine or other unit suitable for use in a computing environment. Also, a computer program can be deployed to be executed on one computer, or to be executed on multiple computers at one site or distributed across multiple sites and interconnected by a communication network.
Certain embodiments are described herein as including one or more modules and optionally, related sub-modules. Such modules are hardware-implemented, and thus include at least one tangible unit capable of performing certain operations and may be configured or arranged in a certain manner. For example, a hardware-implemented module may comprise dedicated circuitry that is permanently configured (e.g., as a special-purpose processor, such as a field-programmable gate array (FPGA) or an application-specific integrated circuit (ASIC)) to perform certain operations. A hardware-implemented module may also comprise programmable circuitry (e.g., as encompassed within a general-purpose processor or other programmable processor) that is temporarily configured by software or firmware to perform certain operations. In some example embodiments, one or more computer systems (e.g., a standalone system, a client and/or server computer system, or a peer-to-peer computer system) or one or more processors may be configured by software (e.g., an application or application portion) as a hardware-implemented module that operates to perform certain operations as described herein.
Accordingly, the term “hardware-implemented module” encompasses a tangible entity, be that an entity that is physically constructed, permanently configured (e.g., hardwired), or temporarily configured (e.g., programmed) to operate in a certain manner and/or to perform certain operations described herein. Considering embodiments in which hardware-implemented modules are temporarily configured (e.g., programmed), each of the hardware-implemented modules need not be configured or instantiated at any one instance in time. For example, where the hardware-implemented modules comprise a general-purpose processor configured using software, the general-purpose processor may be configured as respective different hardware-implemented modules at different times. Software may accordingly configure a processor, for example, to constitute a particular hardware-implemented module at one instance of time and to constitute a different hardware-implemented module at a different instance of time.
Hardware-implemented modules may provide information to, and/or receive information from, other hardware-implemented modules. Accordingly, the described hardware-implemented modules may be regarded as being communicatively coupled. Where multiple of such hardware-implemented modules exist contemporaneously, communications may be achieved through signal transmission (e.g., over appropriate circuits and buses) that connect the hardware-implemented modules. In embodiments in which multiple hardware-implemented modules are configured or instantiated at different times, communications between such hardware-implemented modules may be achieved, for example, through the storage and retrieval of information in memory structures to which the multiple hardware-implemented modules have access. For example, one hardware-implemented module may perform an operation, and may store the output of that operation in a memory device to which it is communicatively coupled. A further hardware-implemented module may then, at a later time, access the memory device to retrieve and process the stored output. Hardware-implemented modules may also initiate communications with input or output devices.
As depicted in
The at least one hardware processor 502 may include, for example, a central processing unit (CPU), a microprocessor, a microcontroller, and/or a digital signal processor (DSP). Further, one or more hardware processors 502 may include one or more execution cores capable of executing instructions and performing operations in parallel with each other.
The one or more data storage devices 504 may include any non-volatile data storage device capable of storing the executable instructions 506 and/or other data generated or employed within the example computing system 500. In some examples, the one or more data storage devices 504 may also include an operating system (OS) that manages the various components of the example computing system 500 and through which application programs or other software may be executed. Thus, in some embodiments, the executable instructions 506 may include instructions of both application programs and the operating system. Examples of the data storage devices 504 may include, but are not limited to, magnetic disk drives, optical disk drives, solid state drives (SSDs), flash drives, and so on, and may include either or both removable data storage media (e.g., Compact Disc Read-Only Memory (CD-ROM), Digital Versatile Disc Read-Only Memory (DVD-ROM), magneto-optical disks, flash drives, and so on) and non-removable data storage media (e.g., internal magnetic hard disks, SSDs, and so on).
The one or more memory devices 508 may include, in some examples, both volatile memory (such as, for example, dynamic random access memory (DRAM), static random access memory (SRAM), and so on), and non-volatile memory (e.g., read-only memory (ROM), flash memory, and the like). In one embodiment, a ROM may be utilized to store a basic input/output system (BIOS) to facilitate communication between an operating system and the various components of the example computing system 500. In some examples, DRAM and/or other rewritable memory devices may be employed to store portions of the executable instructions 506, as well as data accessed via the executable instructions 506, at least on a temporary basis. In some examples, one or more of the memory devices 508 may be located within the same integrated circuits as the one or more hardware processors 502 to facilitate more rapid access to the executable instructions 506 and/or data stored therein.
The one or more data storage devices 504 and/or the one or more memory devices 508 may be referred to as one or more machine-readable media, which may include a single medium or multiple media (e.g., a centralized or distributed database, and/or associated caches and servers) that store the one or more executable instructions 506 or data structures. The term “machine-readable medium” shall also be taken to include any tangible medium that is capable of storing, encoding, or carrying instructions 506 for execution by the machine and that cause the machine to perform any one or more of the methodologies of the present invention, or that is capable of storing, encoding, or carrying data structures utilized by or associated with such instructions 506.
Machine-readable media may also include transitory and non-transitory communication media. Communication media includes computer-readable instructions, data structures, hardware-implemented modules, program modules or other data in a modulated data signal such as a carrier wave or other transport mechanism and includes any information delivery media. The term “modulated data signal” means a signal that has one or more of its characteristics set or changed in such a manner as to encode information in the signal. For example, communication media may include wired media such as a wired network or direct-wired connection and wireless media such as acoustic, RF, infrared, and/or other wireless media, or some combination thereof.
The input/output devices 510 may include one or more communication interface devices 512, human input devices 514, human output devices 516, and environment transducer devices 518. The one or more communication interface devices 512 may be configured to transmit and/or receive information between the example computing system 500 and other machines or devices by way of one or more wired or wireless communication networks or connections. The information may include data that is provided as input to, or generated as output from, the example computing device 500, and/or may include at least a portion of the executable instructions 506. Examples of such network or connections may include, but are not limited to, Universal Serial Bus (USB), Ethernet, Wi-Fi®, Bluetooth®, Near Field Communication (NFC), Long-Term Evolution (LTE), and so on. One or more such communication interface devices 510 may be utilized to communicate one or more other machines, either directly over a point-to-point communication path, over a wide area network (WAN) (e.g., the Internet), over a local area network (WAN), over a cellular (e.g., third generation (3G) or fourth generation (4G)) network, or over another communication means. Further, one or more of one of wireless communication interface devices 512, as well as one or more environment transducer devices 518 described below, may employ an antenna for electromagnetic signal transmission and/or reception. In some examples, an antenna may be employed to receive Global Positioning System (GPS) data to facilitate determination of a location of the machine or another device.
In some embodiments, the one or more human input devices 514 may convert a human-generated signal, such as, for example, human voice, physical movement, physical touch or pressure, and the like, into electrical signals as input data for the example computing system 500. The human input devices 514 may include, for example, a keyboard, a mouse, a joystick, a camera, a microphone, a touch-sensitive display screen (“touchscreen”), a positional sensor, an orientation sensor, a gravitational sensor, an inertial sensor, an accelerometer, and/or the like.
The human output devices 516 may convert electrical signals into signals that may be sensed as output by a human, such as sound, light, and/or touch. The human output devices 516 may include, for example, a display monitor or touchscreen, a speaker, a tactile and/or haptic output device, and/or so on.
The one or more environment transducer devices 518 may include a device that converts one form of energy or signal into another, such as from an electrical signal generated within the example computing system 500 to another type of signal, and/or vice-versa. Further, the transducers 518 may be incorporated within the computing system 500, as illustrated in
The present disclosure provides methods of using a model of the in vivo metabolism of a CNS biomolecule or Aβ. The model may be used to calculate metabolic parameters, such as the synthesis and clearance rates within the CNS, in one aspect. In an aspect, the kinetic model may be used to identify the disease state of a patient by comparing an index calculated from model parameters to a pre-selected threshold. In another aspect, the kinetic model may be used to predict the metabolism and/or concentration of Aβ or its various isoforms in a patient in vivo. In an aspect, the model may be used to create a curve fit for each Aβ isoform time course in a patient. In yet another aspect, the model may be used to identify sensitive pathway components to help design drugs or understand a CNS disease. In even another aspect, the model may be used to investigate changes in the kinetics of the isoforms that may be induced by investigational drugs. In one aspect, the model may be used to characterize Aβ in various patients.
In an aspect, the kinetic model may represent enrichment kinetics of Aβ42 and at least one other Aβ isoform. In this aspect, the labeled moiety may be labeled plasma leucine. One of skill in the art will appreciate that other Aβ isoforms may include, but are not limited to, Aβ37, Aβ38, Aβ39, Aβ40, Aβ41, total Aβ, as well as enzymatic digestion products thereof. The Aβ enrichment kinetics data from a patient may be obtained by the SILK method and may include time course data for Aβ42, Aβ40, and/or Aβ38 in the CSF. In an aspect, the data input into the kinetic model may include the time course of Aβ42 in the CSF and the time course of Aβ40 in the CSF. In another aspect, the data input into the kinetic model may include the time course of Aβ42 in the CSF and the time course of Aβ38 in the CSF. In yet another aspect, the data input into the kinetic model may include the time course of Aβ42 in the CSF, the time course of Aβ40 in the CSF, and the time course of Aβ38 in the CSF. In an aspect, the time course data of labeled plasma leucine may be input into the kinetic model.
The input of the data into the kinetic model may create a set of model parameters for that patient. The model parameters obtained from the kinetic model may include, but are not limited to, the concentration of Aβ isoforms, rates of transfer (e.g. kAPP, kC99, kAb42, kAb40, kAb38), rates of irreversible loss (e.g. vAPP, vC99, v42, v40, v38), rates of exchange (e.g. kex42, kret), rates of delay (e.g. kdelay), or any parameter that may be used in the kinetic model. The model index may be calculated using at least one model parameter. The model index may be calculated using any mathematical operator with the at least one model parameters, including but not limited to multiplication, division, addition, subtraction, logarithm, or any other mathematical operator. In an aspect, the model index may be calculated using the model parameters for the rate of irreversible loss of Aβ42 and the rate of transfer of Aβ42. In one aspect, the model index may be calculated using the calculation shown in Eqn. (I) below:
(10λkAb42)+v42 Eqn. (I)
In another aspect, the model index may be calculated using a model parameter of Aβ42 and the same model parameter of another Aβ isoform (e.g. Aβ37, Aβ38, Aβ39, Aβ40, Aβ41, total Aβ, or other Aβ isoforms known in the art). By way of non-limiting examples, a model index may be calculated using a calculation shown in Eqn. (II) to Eqn. (XI) below:
Aβ42 peak time/Aβ40 peak time Eqn. (II)
Aβ42 peak time/Aβ39 peak time Eqn. (III)
Aβ42 peak time/Aβ38 peak time Eqn. (IV)
Aβ42 peak time/Aβ37 peak time Eqn. (V)
Aβ42 peak time/total Aβ peak time Eqn. (VI)
Aβ42 FTR/Aβ40 FTR Eqn. (VII)
Aβ42 FTR/Aβ39 FTR Eqn. (VIII)
Aβ42 FTR/Aβ38 FTR Eqn. (IX)
Aβ42 FTR/Aβ37 FTR Eqn. (X)
Aβ42 FTR/total AβFTR Eqn. (XI)
Other aspects describing alternative model indices are described herein below in the Examples.
A pre-selected threshold may be calculated in the same manner as the model index using the model parameters of other patients or an average of model parameters from other patients with a known disease state. The method of using the kinetic model to identify the disease state of a patient may include identifying Alzheimer's disease in the patient. In an aspect, the disease state may be identified as Alzheimer's if the model index is above a pre-selected threshold for Alzheimer's. In another aspect, the severity of the disease state may be identified by comparing the model index to a pre-selected correlation of the disease state. In one aspect, the correlation of the disease state may be identified by PIB imaging.
The kinetic model may be used to create a curve fit for each Aβ isoform time course in a patient. In an aspect, limited data from a patient may be input into the model and the model may produce a curve fit for each Aβ isoform time course from the data provided. The curve fit may be used to predict unknown metabolism of Aβ and project to a later time course.
The kinetic model may be used to predict the metabolism and/or concentration of Aβ in a patient. In an aspect, a database of parameters, as described herein above, may be used within the model to predict the metabolism of a Aβ isoform in a patient by using the set of parameters from the database that most closely match the genotype or phenotype of the patient. In another aspect, the model may be used to predict the concentration of different Aβ isoforms at different locations within the body and/or at different time points. In another aspect, the model may be used to calculate the metabolic parameter within the model.
The kinetic model may be used to identify sensitive pathway components to help design drugs or understand a CNS disease. In an aspect, compartments may be added or subtracted to observe the effect of the concentrations and rate constants of the Aβ isoforms. In one aspect, the addition or subtraction of compartments may indicate sensitive areas within the pathway and may indicate areas for potential drug action. In another aspect, the rate constants within the model may be increased or decreased to observe the effect of the concentrations and other rate constants of the Aβ isoforms. In one aspect, the adjustment of the rate constants may indicate sensitive areas within the pathway and may indicate areas for potential drug action.
The model may be manipulated to simulate the action of a drug within the CNS. In an aspect, the model may be used to investigate changes in the kinetics of the Aβ isoforms that may be induced by investigational drugs. In one aspect, the model parameters may be adjusted to best represent the effect of a drug on a patient in vivo. In another aspect, the model may be used to predict CSF concentrations of at least one Aβ isoform CSF concentration.
The model may be used to characterize Aβ kinetics in various patients. In an aspect, the parameters in the database may be used to predict the kinetics of Aβ in other patients. In an aspect, a non-carrier patient may be modeled using the parameters in the database for a non-carrier without the need to measure the concentration of the Aβ isoforms in the CSF. In an aspect, a MC PIB− patient may be modeled using the parameters in the database for MC PIB− without the need to measure the concentration of the Aβ isoforms in the CSF. In an aspect, a MC PIB+ patient may be modeled using the parameters in the database for MC PIB+ without the need to measure the concentration of the Aβ isoforms in the CSF.
The following examples are included to demonstrate preferred embodiments of the invention. It should be appreciated by those of skill in the art that the techniques disclosed in the examples that follow represent techniques discovered by the inventors to function well in the practice of the invention, and thus can be considered to constitute preferred modes for its practice. However, those of skill in the art should, in light of the present disclosure, appreciate that many changes can be made in the specific embodiments which are disclosed and still obtain a like or similar result without departing from the spirit and scope of the invention.
The following experiment assessed the development of a model of Aβ trafficking in vivo using data from SILK studies.
The model consisted of the following structure and parameters. The rate of production of APP was governed by the product of the zero-order rate constant kAPP and the fraction of isotope-labeled leucine. The units of ‘concentrations’ were ng per mL of CSF, thus not accounting explicitly for the volume of the brain compartment. The APP degradation product C99 was produced at a rate governed by the product of the rate constant kC99 and the concentration of APP. C99 was further processed into the three Aβ peptides, Aβ38, Aβ40 and Aβ42 at rates governed by the product of the concentration of C99 and the rate constants kAβ38, kAβ40 and kAβ42, respectively. C99 may also be irreversibly degraded to produce other products, governed by the product of the rate constant {dot over (V)}C99 and the C99 concentration. All irreversible clearance processes that occur within the brain (degradation, transport to the vasculature and deposition into plaques) may be described by product of the rate constants {dot over (V)}38, {dot over (V)}40 or {dot over (V)}42 multiplied by the soluble brain concentration of Aβ38, Aβ40 and Aβ42, respectively. Transport of the CSF to the lumbar space may be modeled as three CSF delay compartments with equal rate constants for entry and exit (kdelay). The concentration of predicted labeled Aβ peptide in the third delay compartment was compared to the total measured concentration of Aβ peptide in the CSF to compute a predicted fractional labeling. The parameters were optimized against the measured fractional labeling of the Aβ peptide.
In vivo SILK studies were performed in participants with ADAD mutations and sibling non-carrier controls. The Aβ kinetic parameters were compared by the presence of a PSEN mutation and insoluble amyloid deposition as measured by PiB-PET.
SILK studies were performed in 23 patients (11 with mutations in PSEN1 or PSEN2, 12 non-mutation carrier sibling controls) using a 9-h primed constant infusion of 13C6 leucine. Seven mutation carriers had evidence of plaques by PiB PET; the remaining mutation carriers and all non-carriers were PiB negative. Four mutation carriers were cognitively symptomatic, all other participants were cognitively normal. CSF Aβ38, Aβ40, and Aβ42 concentrations and isotopic enrichments were measured at hourly intervals over a 36 h period.
During the 13C6-leucine infusion, plasma leucine enrichment approximated a constant plateau and then rapidly decreased after the infusion was stopped (
To compare Aβ isoform kinetics, ratios of labeled Aβ isoform enrichments in the CSF were plotted so that a ratio of one indicates the same isotopic enrichment and kinetics between Aβ isoforms. The Aβ38:Aβ40 labeling ratio was approximately constant at one over time in all patient groups (
A comprehensive compartmental model similar to the models described previously herein was developed to quantify steady state Aβ isoform kinetic parameters. The model incorporated the plasma leucine and Aβ enrichment time course profiles and the CSF Aβ isoform concentrations for each patient (schematic diagram in
The results of this experiment demonstrated that biological mechanisms and patient data that account for Aβ isoform-specific differences may be used to develop a model of Aβ isoform kinetics and the model may provide insights into the metabolic kinetics of Aβ peptides by both mutation and amyloid deposition status.
To demonstrate the ability of the model to account for exchange with unlabeled Aβ peptides, the following experiment was conducted.
Using the model developed in Example 1, additional compartments were added to further develop the model. To optimally fit the shape and peak magnitude of Aβ isoform enrichment time courses, a compartment was required to model reversible exchange of newly synthesized labeled Aβ peptides with a pre-existing pool of unlabeled Aβ, as shown in
9510−5
The results of this experiment demonstrated that a compartment for the exchange of labeled Aβ peptides with unlabeled peptides was necessary to model the exchange of Aβ42, particularly in mutation carrier groups.
To assess the ability of the model to account for irreversible loss, the following experiment was conducted. Using the model of Examples 1 and 2, additional compartments were added to further develop the model. The fractional turnover rate (FTR, pools/h) of soluble Aβ is the rate constant for permanent loss of soluble Aβ and is kinetically distinct from reversible exchange. The physiology of the system suggests that FTR includes irreversible losses to the CSF or bloodstream, degradation, and deposition into amyloid plaques, as illustrated in
The measured concentration of CSF Aβ isoforms were compared by mutation status and PIB score (Table 2). The Aβ42 CSF concentration and the Aβ42:Aβ40 CSF concentration ratio were significantly reduced in association with amyloid deposition (p=0.003 for PIB score; not significant by mutation status), whereas there were no differences between groups for the CSF Aβ38, Aβ40, or Aβ38:Aβ40 concentration ratio. The results of this experiment confirmed that the model may be adapted to account for irreversible loss of each isoform.
To assess the feasibility of using a one dimensional flow model to describe isotope labeling kinetics, the following experiments were conducted.
A one-dimensional flow of Aβ from the brain's interstitial fluid (ISF) to the CSF, was incorporated into a model similar to the mode described in Examples 1 and 2. The model is summarized in the schematic in
All three peptides flow with the brain interstitial fluid and any Aβ peptide that is transported to the surface of the brain without being cleared then becomes part of the CSF. Aβ42 may also enter a reversible exchange compartment, which was previously found to be more substantially exchanged than Aβ38 or Aβ40. Aβ42 within the exchange compartment is not subject to flow. The soluble Aβ42 concentration in the brain does not include the amount of Aβ42 within the exchange compartment. Transport of the CSF to the lumbar space was modeled as two delay compartments with equal rate constants for entry and exit (kdelay).
A length from ventricle to brain surface was taken as 3 cm or 7 cm. The 7 cm value had been adopted in a previous model of ISF flow, but the 3 cm was considered more realistic. The one-dimensional flow model is further summarized in
To develop the one-dimensional flow model from the ventricle to the surface of the brain, the continuity equation was used as shown in Eqn. (2-1) and the one dimensional momentum balance was used as shown in Eqn. (2-2):
where Fv is the rate or production of fluid by the capillaries per unit volume of fluid, v is the velocity of the fluid, and x is the normalized distance from the ventricles. Eqn. (I) expresses the change in velocity of the fluid as due solely to the introduction of new fluid from the capillaries. As more fluid is added, the velocity of the fluid must increase due to the incompressibility of water.
The introduction of fluid from the capillaries due to higher pressure in the vasculature is assumed to follow Starling's Law, as shown in Eqn. (2-3):
F
v
=L
p(S/V)[ρvascular−ρi−(πvascular−πi] Eqn. (2-3)
A non-dimensionalized continuity equation for generality becomes:
where Lp is the hydraulic conductivity, S/V is the surface area of the capillaries per volume of the brain, a is the reflection coefficient, P and IF are pressures, vs is the velocity at the brain surface, and Pi is represented by Eqn. (2-5) below:
After non-dimensionalizing the momentum balance equation and ignoring higher order terms, the momentum equation reduces to Eqn. (2-6) (Arifin et al, 2009, Pharma. Research, 26:2289):
Combining the dimensionless continuity and momentum equations reduces to Eqn. (2-7):
where α and β may be represented by Eqns. (2-8) and (2-9):
This shows that the flow is pressure driven without substantial viscous losses other than due to the porosity alone. The velocity may be calculated from the now-known pressure profile:
where A and B may be represented by Eqns. (2-11) and (2-12):
Using Eqns. (2-7) and (2-10), the pressure and velocity have the profiles across the brain shown in
For transport of an Aβ peptide, the mass balance is (neglecting diffusion due to the P):
where kC99 is the rate of creation of C99 and {dot over (V)}Aβ is the rate of irreversible loss of Aβ.
After partially non-dimensionalizing the steady state equation, the effects of diffusion and time dependent terms may be neglected without introducing substantial error, resulting in the steady state equation below:
The expression for velocity as a function of x calculated in Eqn. (XI) was inserted into Eqn. (2-15) and integrated with a boundary condition of Aβ(0)=0 which yielded the Aβ steady state equation below:
The ‘brain’ was divided into 100 equally spaced nodes and the unsteady system of differential equations was solved numerically for iAPP, mAPP, C99 immobilized in the brain, Aβ38, Aβ40, and Aβ42 in the interstitial fluid and CSF, and Aβ42 in the exchange compartment (705 equations).
The results of this experiment demonstrated that the one-dimensional flow of Aβ in the brain may be modeled.
To assess the model of one-dimensional flow of Aβ in the brain, the following experiment was performed. The model of Examples 1 and 4 were used to model patients that were normal controls (NC) PSEN1 or PSEN2 mutation carriers that were both PIB positive (MC+) and negative (MC−).
The rate of production of labeled iAPP was the product of the rate constant kiAPP with the fractional labeling of leucine amino acid. This value was set to 25 h−1 for all patients. For the production rate constant of mAPP and C99, different values were investigated while fitting the data to one of the patients with plaques detectable by PET. In Example 2, six out of seven of the patients with plaques required exchange of Aβ42 to optimally fit the data, and many required a large amount of exchange. This is due to the characteristic shape of the curves (
The ratio of the rate constant for the production of Aβ42 with respect to the rate constant for the production of Aβ40 was highly significant when comparing both the MC− and MC+ groups to the normal controls (NC). However, the MC− and MC+ groups were not different from each other.
The ratio of the rate constant for the permanent loss of Aβ42 ({dot over (V)}42) with respect to the rate constant for the permanent loss of Aβ40 ({dot over (V)}40) was also highly significant when comparing both the MC− and MC+ groups to the normal controls. Although it was expected that only the MC+ group should show increased loss of Aβ42 relative to Aβ40, it is possible that some patients in the MC− group were beginning to deposit plaques, but these were not yet detectable by PIB. This is supported by the significant increase in the exchange rate constant in the MC− group (p=0.19), although the mean was nearly four-fold smaller than in the MC+ group. The rate constant for permanent loss was 33% higher in the MC+ compared the MC− group, and this difference trended towards significance (p=0.057). Interestingly, the increased {dot over (V)}42/{dot over (V)}40 ratio in the MC+ group seemed to be due to a significant decrease in {dot over (V)}40 rather than an increase in {dot over (V)}42. This result is in agreement with the findings of the purely compartmental model of the data in Examples 1 and 2. However, in this model, the rate constant for the clearance of Aβ38 ({dot over (V)}38) is also significantly lower in the MC+ group. This may represent a general decrease in clearance of from the brain in the presence of plaques, perhaps due to changes in the physiology of the brain.
Compared to the model in Examples 1 and 2, the AIC was lower in the Example 2 model in 13/23 patients, and was lower in the current model in 10/23 patients. However, the AIC were quite similar, with the sum of AIC over all the patients of −25,818.2 for the previous model and −25,729.7 for the current model.
In the current model, only exchange of Aβ42 is allowed, and this parameter is allowed to vary in all patients. In the Example 2 model, patients were allowed to exchange Aβ peptides only if it improved the AIC. The current model treats the exchange rate constant as a continuous variable, this facilitates comparison of this parameter with other measures of Alzheimer's disease. In particular, the correlation between the exchange rate constant and the PIB score is presented. The correlation coefficient of r=0.851 indicates high correlation between the two measures. In contrast, the correlation coefficient between the predicted brain pool size of Aβ42 and the exchange rate constant was r=−0.441. This indicates some relationship between these variables.
The results of this experiment demonstrate that the model may represent one-dimensional flow of Aβ in the brain.
In one embodiment, the computing device 102 or client 108 executes the MCA 104 in response to a modeling request from the user. The user identifies one or more patients for whom Aβ modeling will be calibrated using the input device 120 and one or more GUI's generated by the GUI module 300.
A GUI module 300 receives data from the various other modules 302-310, the input device 120, and/or the data source 106 and generates one or more displays on the display device 116. The displays generated by the GUI module may include input forms, charts, graphs, displays, tables, and other data for viewing by the user of the MCS 100.
In response, the patient data module 302 generates a request to retrieve patient data. In one embodiment, the request is transmitted to the data source 106 to retrieve patient data. The patient data may include biographical data as well as medical data for the identified patient. The patient data may also identify a diseased state of a patient. The patient data may further include baseline data values related to one or more component levels within the patient's blood, CSF, or other baseline data of interest. Alternately, if the MCA 104 is being executed contemporaneously with a new patient, the request for patient data may be transmitted to the GUI module 302, where one or more GUI's and data entry fields are generated for display on the display device 116 for the user to input baseline values, which are received at the patient data module 302.
Once baseline values for the patient have been established, the MCA 104 determines a plasma leucine enrichment value for the patient. The plasma leucine enrichment value is calculated by referencing known data enrichment values as a function of time, as shown in
As previously described, a time-dependent delay compartment of the model is used to represent the uptake of the labeled plasma leucine by APP and the subsequent formation of the Aβ isoforms by cleaving C99 peptides. As such, the MCA 104 includes an Aβ isoforms module 304 that determines the level of each Aβ isoform after cleavage, which incorporates the labeled leucine. The Aβ isoforms module 304 determines the amounts or values for each labeled isoform as well as each isoform's respective enrichment levels by first multiplying the determined plasma labeled leucine level by an uncalibrated APP constant (kAPP), as identified in Table 1, to obtain an uncalibrated level of enriched C99 peptides. The exemplary uncalibrated APP constant is retrieved from a table of mean data values stored in the data source 106. Similarly, the Aβ isoforms module 304 determines an exemplary level for each Aβ isoform entering the CSF by multiplying the calibrated level of enriched C99 peptides by a mean transfer rate values for each respective isoform cleaved from C99 peptides. This determination also accounts for a certain level of the C99 peptides that are lost and not converted to the Aβ isoforms by using an exemplary irreversible loss C99 constant (Vc99).
In one embodiment, the Aβ isoforms module 304 may also be used to calibrate and quantify the state-state kinetics of isoforms. For example, the model may be used to model the kinetics of the Aβ38, Aβ40, and Aβ42 isoforms.
In one aspect, the Aβ isoforms module 304 may be used to determine if an exchange compartment is necessary to model the kinetics of the “soluble” peptides. The module 304 optimizes the model by creating the exchange compartment in response to a determination that the added exchange process improves the Akaike Information Criteria (AIC) for a curve fit. For example, data from exemplary modeling performed using SAAM II software may be stored in the data source 106. In particular, the user or the MCA 104 may automatically incorporate one or more exchange compartments into the exemplary model to calibrate and improve the correspondence between the sigmoid shapes of the enriched Aβ-isoforms within the CSF with respect to time as compared to data in the data source 106.
When exchange compartments are used, the Aβ isoforms module 304 multiplies the previously calculated isoform levels by an exemplary exchange rate (Kex) and an exemplary return rate (Kret). The exchange compartments and rate factors Kex and Kret are used to represent the possible recycling of Aβ isoforms to and/or from amyloid plaques, the exchange of labeled Aβ for unlabeled Aβ, the recycle of higher order Aβ structures, and other as of yet unknown losses and gains to the levels of the respective isoforms.
In addition, the Aβ isoforms module 304 may multiply the calculated isoform levels by one or more scaling factors to account for small amounts of isotopic dilution between plasma leucine and the biosynthetic precursor pool (generally <5%) or to correct for minor calibration errors (generally <10%) in the measurement of isotope enrichments of plasma leucine and/or Aβ peptides.
The CSF isoform module 306 receives data related to the levels of each respective isoform within the CSF. In one aspect, the CSF isoform module 306 receives data regarding the measured or calculated isoform levels after cleavage from the C99 peptide, and/or levels calculated from one or more optional exchange compartments. In addition, the CSF isoform module 306 may be used to predict the levels of each isoform within the CSF as a function of time by multiplying the received data by an exemplary delay factor (Kdelay). As shown in the kinetic model 20, Kdelay may be used to represent the perfusion of labeled peptides through various brain tissue and heterogeneous CSF fluid transport processes.
The results module 308 processes data transmitted from the data source 106 and/or one or more other modules 300-306, and 310 to generate a display of results generated by the kinetic model 20. In one example, the results module 308 may generate a chart or other graphical representation of data values, while the GUI module 302 generates a display of the representation.
The calibration module 310 allows the user to modify one or more of the rate constants or other constants used in the kinetic model 20. In one aspect, the calibration module 310 in conjunction with the GUI module 300 and/or the results module 308 generates one or more GUIs that a user may interact with to modify the parameters of the model, the data values generated by the model, and/or the graphical representation of the data values. By way of example and not limitation, the calibration module 310 may receive data input into a GUI using the input device 120 to modify a constant value of the kinetic model 20. This input data may be used to modify one or more graphical representations generated by the results module 308. As such, the user may vary the data values generated by the kinetic model 20, which contemporaneously varies the graphical representation of the data in order to calibrate the model data values to the measured data value.
A determination regarding the fit or closeness of fit between the plots of measured data and the plots generated by the model is made at 410. If the model-generated plots are determined to sufficiently fit the plots of measured data, the model may be deemed calibrated and used as a tool in other investigations at 412. Conversely, if the model-generated plot does not fit the plots of measured data, then one or more of the rate constant values may be modified at 414 and the model may be re-executed at 416. Similar to the comparison made at 408, the plot generated by the model using the modified rate constant(s) is compared to the plot of the measured data from 402 at 418. Another determination is made at 410 to determine if the “modified rate constant” plot sufficiently fits the plot of measured data. The process at 410-418 may be repeated as necessary, until the user is satisfied with the calibration of the model. In various embodiments, the same rate constant, different rate constants, or combinations thereof may be modified at 414.
The description above includes example systems, methods, techniques, instruction sequences, and/or computer program products that embody techniques of the present disclosure. However, it is understood that the described disclosure may be practiced without these specific details. In the present disclosure, the methods disclosed may be implemented as sets of instructions or software readable by a device. Further, it is understood that the specific order or hierarchy of steps in the methods disclosed are instances of example approaches. Based upon design preferences, it is understood that the specific order or hierarchy of steps in the method can be rearranged while remaining within the disclosed subject matter. The accompanying method claims present elements of the various steps in a sample order, and are not necessarily meant to be limited to the specific order or hierarchy presented.
It is believed that the present disclosure and many of its attendant advantages will be understood by the foregoing description, and it will be apparent that various changes may be made in the form, construction and arrangement of the components without departing from the disclosed subject matter or without sacrificing all of its material advantages. The form described is merely explanatory, and it is the intention of the following claims to encompass and include such changes.
The kinetic tracer curves for Aβ42 are known to differ compared to other index peptides (for example, Aβ38 and Aβ40) for certain patient populations. The data reflect the involvement of plaques, as evidenced by PIB scores. A compartmental model was developed as one way of extracting kinetic parameters from the experimentally measured data. Numerous models may be used to describe the data, and it is predicted that all such models will reveal differences in Aβ42 kinetics if they provide satisfactory fits to the data. The following summarizes outcomes apparent in the raw data itself that are independent of the type of compartmental or non-compartmental model that might be used to describe the data, and demonstrates that the SILK tracer kinetic protocol reveals differences in Aβ42 kinetics that will be diagnostic of plaques.
The kinetic tracer curve for Aβ42 and the other index peptides (e.g. Aβ38, Aβ40) was different during several different phases of the curve in the presence of plaques.
(i) Fractional Synthesis Rate (Uses 6-12 h TTR Slope and Plasma Leucine TTR Enrichment)
None of the Aβ peptide (ABxx) FSRs discriminate PIB status or correlate with PIB score. The Aβ42/Aβxx ratios are lower in PIB+ group (significant when Aβ38 or Total AB is used for normalization, but not when AB40 is used). The Aβ42/38 FSR ratio is significantly negatively correlated with PIB score, but a P value of 0.028 is not that impressive in comparison to other outcomes (see below).
(ii) Time to Peak
None of the individual Aβ peptide peak times discriminate between PIB groups or are significantly correlated against PIB score. However, Aβ42 peaks significantly earlier than either Aβ38, Aβ40, or total AB in the PIB+, and the ratios of the peak times is very highly significantly correlated with the PIB score. Thus, the degree to which the Aβ42 peak is shifted earlier correlates with plaque involvement.
(iii) Peak Enrichment
In these data, enrichment is measured as tracer-to-tracee ratio, but other units of enrichment could be used instead. By itself, the peak enrichment of Aβ42 discriminates between PIB+/− groups and is significantly negatively correlated with PIB score (higher PIB score=lower peak enrichment); but the P value of 0.016 on this is not strongly significant. The lower Aβ42 enrichment is much more strongly associated with plaques when it is normalized to the other index proteins, either Aβ38, 40, or total Aβ. This normalization is crucial as it controls for variability in the plasma leucine enrichment plateau between subjects that is observed with the SILK protocol.
(iv) Initial Monoexponential Slope FCR
A monoexponential slope is fit to the descending enrichment on the back end of the time course. In most studies, the entire back end of the peak is monoexponential to the end of the time course (36 h) as shown in
None of the individual peptide monoexponential slopes significantly discriminate between PIB groups, although there is a trend that Aβ40 and total Aβ have slower slopes in the PIB+ group. Greater discriminatory power is achieved by looking at the correlation against PIB score, where the monoexponential slopes for Aβ38, Aβ40, and total Aβ are all significantly negatively correlated against PIB score (slower slope in relation to the degree of plaque quantity). In the formal compartmental model, this came out as a decreased fractional turnover rate (FTR) of soluble Aβ38 & Aβ40 in the brain in the presence of plaques.
However, the Aβ42 monoexponential slope does not significantly discriminate between PIB groups nor does it correlate with PIB score. The FTR of soluble Aβ38 and Aβ40 was slowed down in the presence of plaques. This turnover is largely due to fluid perfusion through the brain, and we propose that the fluid perfusion rate is slowed down in the presence of plaques. In the compartmental model, it is assumed that the FTR of Aβ42 that is due to the fluid perfusion process would be the same as it is for Aβ38 and Aβ42. Since the initial monoexponential slope of Aβ42 is not significantly slower in the presence of plaques, but it should be if fluid perfusion was the sole process for Aβ42 turnover, we therefore concluded that some other process of irreversible loss was causing the total FTR of Aβ42 (fluid perfusion loss+extraneous loss) to be increased selectively in the PIB+ group. We take this as kinetic evidence for removal of soluble Aβ42 from the brain fluid and deposition into plaques, which accounts for the observation that the initial monoexponential is not slower in the presence of plaques (even though the slopes of Aβ38 & Aβ40 are slower), and also provides a mechanism that reduces the concentration of AB42 relative to Aβ38 or Aβ40 that is recovered in CSF.
The Aβ42 initial monoexponential slope also fails to discriminate between PIB groups or correlate with PIB score when it is normalized using either Aβ38, Aβ40 or total Aβ as a reference. Thus, in conclusion, the initial monoexponential slope FCR of Aβ42 is not diagnostic of plaques.
(v) Terminal Monoexponential Slope FCR
A monoexponential slope was fit to t=24-36 h of the time course as reported in the Science 2010 paper; this is done without regard for whether the peak exhibits monoexponential or biexponential behavior (see natural log plots in
By itself, the terminal slope of Aβ42 very weakly discriminates between PIB groups (P=0.0355), with PIB+ having a slower terminal tail. In the model, this is accounted for by the “exchange compartment” whereby newly synthesized (i.e., labeled) Aβ42 enters into an exchange process that returns labeled Aβ42 to the soluble pool later, which is a feature of tracer recycling that causes a flattening of the terminal tail. The terminal slopes of Aβ38 or Aβ40 do not discriminate between PIB groups. The terminal slopes of all 3 peptides, however, are significantly negatively correlated with PIB score, which results from the feature described above whereby the turnover of soluble Aβ peptides may be mostly driven by fluid transport through the brain tissue, and this transport process is retarded in the presence of plaques. The small degree of discrimination between groups for Aβ42 is lost when that slope is normalized to either the Aβ38 or Aβ40 slope. In conclusion, the terminal monoexponential slope of Aβ42 is not particularly diagnostic for plaques. There is a weak power to discriminate, but the enrichment measurements are somewhat noisy (especially as enrichments get lower toward the end of the protocol), and the slope is not all that useful.
(vi) Overall Conclusions
The peak time and peak enrichment of Aβ42 is very highly significantly associated with plaques: Aβ42 peaks earlier and lower when plaques are present. The slope on the front end and the initial and terminal monoexponential slopes on the back end are not particularly sensitive to the presence of plaques.
The presence of plaques clearly alters biologic processes that distinguish the Aβ42 turnover curve from Aβ38, Aβ40, or total Aβ. The earlier and lower peak of Aβ42 in the presence of plaques (peak time and peak enrichment, respectively) causes a separation of enrichments on the back end of the curve (see time course plots). In addition to these two measurements, recent results show that a comparison of isotopic enrichments around the midpoint on the back end of the curve (˜24 h) is also able to discriminate the PIB groups highly significantly. A fourth measurement that may be associated with plaques is the degree to which Aβ42 enrichment on the descending peak is different from Aβ38, Aβ40, or Total Aβ enrichment.
It was hypothesized that simple measures that summarize some aspect of the SILK tracer curve of amyloid beta (Aβ) may provide diagnostic or prognostic information about patients with AD, at risk of AD, or suspected of having AD. To test the above hypothesis, discrimination between the three groups of patients was attempted based on the ratio of the percent of Aβ42 labeling to the percent of Aβ40 percent calculated during the downturn of the Aβ SILK tracer curve. In vivo SILK studies were performed in patients with PSEN1 or PSEN2 mutations that were PIB positive by PET (MC+), patients with PSEN1 or PSEN2 mutations that were PIB negative by PET (MC−), and non-carrier mutation carrier sibling controls (NC) as described elsewhere in U.S. Pat. No. 7,892,845, which is hereby incorporated herein in its entirety. Briefly, subjects were administered isotope-labeled leucine (13C6-leucine) for 9 hours via intravenous infusion. CSF samples (6 mL/sample) were collected 23 hours and 24 hours after the start of the infusion of labeled amino acid. Quantitative measurements of labeled and unlabeled Aβ42 and Aβ40 were obtained by tandem mass spectrometry, and the ratio of labeled:unlabeled Aβ42 and labeled:unlabeled Aβ40 was calculated for each timepoint. These ratios represent the percent labeled of each Aβ isoform at 23 hours and 24 hours post infusion.
A diagnostic threshold of 0.9 was defined in these experiments, such that a ratio of Aβ42 percent labeled/Aβ40 percent labeled below 0.9 classified a subject as AD positive and a ratio of Aβ42 percent labeled/Aβ40 percent labeled above 0.9 classified a subject as AD negative. To determine whether the ratio of Aβ42 percent labeled/Aβ40 percent labeled at 23 hrs post infusion was differentiated between the three groups of patients, the ratio obtained for each patient was graphed versus PIB staining. As can be seen in
These data may be compared to a simple measure that uses the results from the full kinetic model. In this case, the parameter kex42, which describes the rate of entry of Aβ42 into the exchange compartment, is multiplied by 10 and then added to the ratio of the rate constants for irreversible loss for Aβ42 versus Aβ40. As shown in
These examples indicate that simple measures that summarize some aspect of the SILK tracer curve may be diagnostic of AD. This also indicates that short term collection of CSF may be sufficient to diagnose changes in Aβ42 kinetics.
The Aβ precursor protein (APP), produced in high amounts by neurons, is known to be degraded by different enzymes [2]. The enzyme β-secretase cleaves APP to produce the C99 peptide. C99 is then further processed by α-secretase to produce Aβ peptides of different lengths (e.g. Aβ38, Aβ40, Aβ42, where the number indicates the number of amino acids in the peptide). Aβ peptides are able to self-aggregate, with Aβ42 being more prone to formation of large aggregates [3], and the major constituent of senile plaques [4].
A promising approach to characterize the kinetics of Aβ production and clearance in humans relies on in vivo labeling of Aβ peptides during protein translation via infusion of stable isotope-labeled amino acids, stable isotope labeling kinetics (SILK) [5]. The fraction of isotope-labeled Aβ is measured at timed intervals in cerebrospinal fluid (CSF) collected at the lumbar subarachnoid space. The traditional method to estimate rates of irreversible loss of Aβ peptides from the CNS is analysis of the terminal slopes of isotopic enrichment time course curves evaluated on log-normal plots. This analysis method yields a measure that is referred to herein as the monoexponential fractional clearance rate (monoexponential FCR) [6]. Previous results demonstrated decreased monoexponential FCR of both Aβ40 and Aβ42 in late-onset AD [7]. However, the monoexponential FCR should not be confused with the true underlying fractional clearance rate, which may be difficult to determine in complicated systems. The true fractional clearance rate is the rate of irreversible loss of a product divided by the pool size of the product. To avoid confusion, the term fractional turnover rate or FTR is used herein, which has the same meaning as the true fractional clearance rate. The FTR is also equal to the sum of all of the rate constants describing routes of irreversible loss. The fractional synthesis rate (FSR) was determined by fitting a line to the upslope of the isotopic enrichment time course curve. FSR is defined as “the rate of incorporation from precursor to product divided by the pool size of the product”[6]. Thus, FSR is distinct from the production rate constant, which is the rate of incorporation from precursor to product divided by the pool size of the precursor. The FSR and monoexponential FCR analysis methods were acknowledged to have limitations, in that they imposed a simple one-compartment model on a complicated system [8]. However, more physiologically relevant models had not yet been developed.
Examples 1-8 introduced a physiologically relevant multi-compartmental model to distinguish carriers of presenilin-1 or presenilin-2 mutations that are the active enzymatic components of α-secretase and result in onset of AD at younger ages than non-mutation carriers (familial autosomal dominant AD). This work is also presented in detail in Sci. Transl. Med. 2013, pp. 189ra77, which is hereby incorporated by reference in its entirety. The main strength of the new model is that the rates of production, transport, reversible and irreversible loss of APP, C99, and the Aβ peptides may be estimated by fitting the model to the entire time course of the isotopic enrichment data while also accounting for the Aβ peptide concentrations in CSF. The model successfully detected an increase in the rate of production of Aβ42 relative to Aβ40 in human subjects with presenilin mutations, consistent with results in vitro and in mice [10]. Increased FTR of soluble Aβ42 relative to Aβ40 were also detected in participants known to have senile plaques demonstrated by positron emission tomography (PET) using Pittsburgh compound B (PIB). The previous observation of decreased monoexponential FCR of Aβ42 in late onset AD was re-interpreted in the context of amyloid positive mutation carriers when the full enrichment time courses were fit to the compartmental model [7]. From the analysis of Aβ isoforms in mutation carriers, it was concluded that the data actually reflected increased irreversible loss of soluble Aβ42 relative to Aβ40. Faster irreversible loss in combination with exchange of Aβ42 with higher order structures (e.g. aggregates, micelles, or the surface of pre-existing plaques) resulted in a ‘slower’ terminal exponential tail.
The compartmental model answered several questions concerning the amyloid hypothesis. However, the previous publication on the compartmental model did not discuss the identifiability of particular parameters[11] and [12]. In Examples 11-16, the identifiability of the different parameters in the compartmental model is described via a parameter sensitivity analysis. Analysis of the steady state of the model also revealed a potential mechanism for the decrease in the CSF concentration of Aβ42 in Alzheimer's disease [13].
Examples 11-16 refer to appendices A-K. Appendices A-K are Supplementary Materials to the publication entitled “Analysis of a compartmental model of amyloid beta production, irreversible loss and exchange in humans” (Mathematical Biosciences, 2015, pp. 48-61, Vol. 261). The publication and its Supplementary Material are incorporated herein by reference in their entirety.
Experimental methods for isotopic labeling of Aβ peptides and measurement of their concentrations in CSF are described in a separate publication [9]. Systems identifiability analysis and sensitivity analysis were performed as described in the text.
A compartmental model was constructed to describe Aβ peptide-labeling data
Prior to the addition of labeled leucine, a steady state was presumed whereby the rate of production of the unlabeled protein (kAPP) was equal to the rate of conversion to C99 (−kC99×cAPP) or other products (−vAPP×cAPP). A steady state pool size of APP (concentration of APP multiplied by compartment volume) was assumed throughout the labeling experiment, thus:
To simplify the analysis, the fraction of isotopically labeled leucine in plasma was taken to be the average value during the labeling phase, f. Rates of change in the pool size of unlabeled APP (cAPP) and labeled APP (cAPPL) during the infusion phase are thus:
These equations imply that labeled leucine is added to tRNA proportional to the fraction f of labeled leucine, not the tracer-to-tracee ratio (TTR). The fraction f is [labeled leucine]/([unlabeled leucine]+[labeled leucine]), while the TTR is [labeled leucine]/[unlabeled leucine]. The fraction f has been shown to be the appropriate model for protein synthesis [15], and differs from the previous analysis method that used the TTR [7], although at the limit of low enrichment this is a minor difference. The use of TTR versus fractional labeling is further described in Appendix C.
The equations are solved:
The initial conditions at the moment of addition of labeled amino acid were:
The solutions appropriate for these initial conditions are:
If the rates of production and irreversible loss of APP do not change during the course of the labeling experiment, the pool sizes of labeled plus unlabeled protein will equal the original steady state pool size of protein:
c
APP, SS
=c
APP
+c
APPL Eqn. (11.1.11).
Stated another way, the pool size of unlabeled protein must decline because a fraction of the tRNAs are loaded with the labeled amino acid.
The fractional labeling of APP (pAPPL) is obtained by dividing Eqn. (11.1.10) by Eqn. (11.1.2):
The same result would be obtained by dividing the rate equation for cAPPL by the steady state concentration of APP and solving this differential equation:
Notice that the ‘rate of appearance’ of labeled APP in Eqn. (11.1.12) does not depend on the parameter kAPP. Basic kinetic intuition would suggest that the slope of the initial portion of the labeling curve should equal to the APP synthesis rate constant kAPP. This would be true if concentrations or pool sizes were measured, but not if TTR or fractional labeling is measured. To see this, the exponential terms in the equations for cAPPL and pAPPL are expanded as Taylor series in time. Assuming very short times, the terms in t2 and higher may be neglected:
However, for fractional labeling:
Although this section specifically described APP production and clearance rates, the conclusions are valid for any one compartment model. The initial slope of a labeling curve for a one-compartment model yields a measure of the irreversible loss rate constant and not its production rate constant. However, later it will be shown that the upslope of a labeling curve for a system described by a multicompartment model is more complicated.
At the end of the labeling period, the infusion of labeled amino acid ceases. The labeled fraction of isotope-labeled leucine in plasma drops rapidly and is well-described by a bi-exponential decay:
f=f
0(αe−qmt+βe−qrt) Eqn. (11.2.1)
where f0 is the fraction of labeled amino acid in plasma during the labeling period. For compactness, t=0 in this equation corresponds to the end of the labeling period. The sum of the parameters α and β is one. The parameters α and qm tend to be large and presumably represent rapid clearance of the labeled amino acid throughout the body. The parameters and qr tend to be much smaller and likely represent reappearance of labeled leucine in plasma due to exchange of labeled plasma amino acid with non-plasma spaces and/or incorporation into and subsequent degradation of rapidly turning over proteins throughout the body.
At the end of the labeling period, labeled APP had a pool size of cAPPL, end. The rate equation for labeled APP Eqn. (11.1.4) is solved with the new expression for f and with initial condition cAPPL(0)=cAPPL, end:
The pool size of APP at the end of the labeling period is obtained from Eqn. (11.1.10):
where tend is the length of the labeling period.
Dividing by the pool size of APP at steady state, the fractional labeling of APP after removal of labeled amino acid is:
The first term on the right hand side represents new synthesis of labeled APP due to residual labeled amino acid. If the ‘new synthesis’ term is neglected, then a semilog-y plot would yield:
ln(pAPPL)=ln(pAPPL,end)−(kC99±vAPP)t Eqn. (11.2.5)
with slope of −(kC99+vAPP). This illustrates the fact that the downslope of a one compartment model would yield an approximation of the rate constants describing ‘clearance’ but no information about rate constants of ‘production’ (the full model described below does not neglect new synthesis, unlike simple fits of monoexponential curves to the downslope of the labeling curve).
The rate equation that describes production of labeled C99 during the labeling phase is:
The rate constants kAb38, kAb40, and kAb42 govern the rate of production of Aβ38, Aβ40 and Aβ42, respectively. The rate constant vC99 describes all other irreversible losses of C99, including production of Aβ peptides of other molecular weights. For compactness, kAb=kAb38+kAb40+kAb42+vC99.
The rate equations for all three Aβ peptides are similar and will be elaborated for Aβ42 only. The rate equation for labeling kinetics of soluble Aβ42 in the ‘brain’ is:
The parameter kAb42 represents production of Aβ42 from C99 by the action of α-secretase, v42 describes irreversible loss of Aβ42 from the soluble brain compartment by means other than transfer to CSF, kCSF describes irreversible loss into CSF, kex42 describes entry into an exchange compartment, kret42 describes return of Aβ42 from the exchange compartment to the ‘brain’ compartment, and cAβ42exL is the pool size of labeled Aβ42 in the exchange compartment.
The kinetics of entry/exit of labeled Aβ42 into/from the exchange compartment is described by the rate equation:
Without wishing to be bound by theory, it is believed that the ‘exchange’ compartment represents a reversible interaction with higher order structures, perhaps with the surface of amyloid plaques or oligomers (see reference [9] and Appendix A for additional discussion) [9]. In contrast, permanent or even slowly reversible assimilation into stable plaques would lead to an increase in the parameter v42, because the labeled Aβ would not return to the soluble form during the time course of the experiment. This would thus be indistinguishable from other mechanisms of irreversible loss of Aβ42.
The rate equations for the three CSF delay compartments are:
1) First CSF delay compartment:
2) Second CSF delay compartment:
3) Third CSF delay compartment:
The system of differential equations in terms of fractional labeling may be written as Eqn (11.3.7), shown in
Eqn. (11.3.8) describes the shape of the isotopic enrichment time course curve according to this compartmental model. An important conclusion is that the rate constant for production of Aβ42 (kAb42) does not appear in these equations except through its inclusion in kAb. Thus, any impact that kAb42 has on Aβ labeling kinetics would only be manifest if this caused an increase in the rate of turnover of C99. If increases in secretase activity to produce Aβ42 are exactly balanced by decreases in production of other Aβ isoforms, then the model predicts that increases in the rate of production of Aβ42 would not be detectable by an isotope labeling experiment alone. However, as will be shown by a steady state analysis, the rate of production of Aβ peptides may be calculated by using both Aβ42 isotope labeling kinetics and CSF Aβ42 concentration data.
The goal of the experimental studies was to determine the rate constants for the production (kAb38, kAb40 and kAb42) and irreversible loss (v38, v40 and v42) of the Aβ peptides. In the former case, this may sometimes be stated as determining the ‘production rates of the Aβ peptides’. Because the Aβ peptides have a common precursor (C99), the production rate constants are in fact the true determinants of the production rates. Similarly, it may be stated that the ‘clearance rates’ are of interest. However, this is much less precise, because these rates (or fluxes, both with units of mass/time or concentration/time) depend on the pool size/concentration of each Aβ peptide, which differ greatly. In fact, the kinetic measures that allow meaningful comparisons of irreversible loss between the different Aβ isoforms are the ‘clearance rate constants’ or ‘irreversible loss rate constants’.
Because the models are at steady state, the production rate and irreversible loss rate must be equal. Thus, only one rate is required, the ‘turnover’ rate [16] and [17]. The turnover rate divided by the concentration or pool size of the product is the fractional turnover rate (FTR), which is equal to the irreversible loss rate constant. The ‘fractional synthesis rate’ is the rate of appearance of labeled product divided by the pool size or concentration of the product [6], which is the same as dividing the turnover rate by the product pool size. Thus, the fractional synthesis rate is theoretically the same as the FTR (i.e. true FCR) and the irreversible loss rate constant. However, the ‘FSR’ often refers to the method of estimating the fractional turnover rate by fitting a line to the upslope of a curve and dividing the slope by the enrichment of the precursor. This method of estimating FTR is only accurate for systems well-described by single-compartment models. However, CSF Aβ kinetics are best described by a multi-compartmental model, and the ‘FSR’ that was previously applied to CSF Aβ kinetics [7] and [14] may thus actually reflect changes in the production rate constant, one of the two quantities of interest.
where cproductx, SS is the steady state concentration of product x.
In
Changes in k1 or k2 also do not impact the labeling curve as long as +k2 remains constant (
In addition to measurement of the fractional labeling of each of the Aβ peptides in the CSF, the concentration of each peptide in the CSF was measured by mass spectrometry. The concentrations of the Aβ peptides in CSF provided additional constraints on the parameters in the system. Although some diurnal variation in Aβ42 concentration in the CSF has been noted [18], the concentration in CSF at the start of the experiment was assumed to represent a steady state throughout the experiment.
To calculate pool size in each CSF compartment, the measured CSF concentration was multiplied by a typical CSF volume of 135 mL, and divided by 3 to account for three equal-volume CSF compartments in the model. The assumption that every participant had a CSF volume of 135 mL divided into three compartments seems strong but actually has little impact on the results. If the CSF concentrations of Aβ peptides were used instead of pool sizes, the results for all of the first-order rate constants would be identical, but the zero-order rate constant kAPP would simply be lower by a factor of 3/135. Because kAPP does not affect the shape of the predicted isotope-labeling curve, use of either concentrations or pool sizes in fitting the labeling curves is justified.
According to the current model, the pool size of Aβ42 measured in the lumbar CSF is equal to the steady state pool size of Aβ42 in the third delay compartment (see Eqn. 11.3.5 and 11.3.6). The steady state pool sizes of Aβ42 in each of the three delay compartments must be equal:
C
Ab42,delay3,SS
=C
Ab42,delay2,SS
=C
Ab42,delay1,SS Eqn. (11.5.1).
Relative to the pool size of soluble Aβ42 in the brain, the pool size of Aβ42 in each delay compartment is predicted to be scaled by a factor kCSF/kdel,
The exchange compartment has no effect on the steady state pool size of soluble Aβ peptides in the brain or CSF. However, the pool size of the exchange compartment itself is:
Deposition into plaques or aggregates that do not return labeled Aβ42 on the time scale of the experiment would only impact the irreversible loss parameter, v42. Thus, rates of deposition of Aβ42 into plaques can be estimated by comparing the difference between v42 and v40, or between v42 and v38, because Aβ38 and Aβ40 deposition into plaques is expected to be minimal [19].
After additional substitutions for the steady state concentrations of APP, C99, and Aβ peptides in the brain, the steady state concentrations in the CSF for each of the Aβ peptides is predicted to be:
Overall, the model has 25 parameters:
kAPP, vAPP, kC99, cC99, kAb38, kAb40, kAb42, V33, v40, v42, kex38, kex40, kex42, kret38, kret40, kret42, kCSF38, kCSF40, kCSF42, kdel38, kdel40, kdel42, SF38, SF40, SF42
The last three parameters are scaling factors that were applied to the predicted labeling curve for each peptide. The scaling factors were found to improve the fit and may correct for systematic errors caused by variability in the standard curves used in the daily calibration of the mass spectrometers, or isotopic dilution between plasma leucine and APP production. The mean values of the scaling factors for all participants were 0.941±0.08, 0.944±0.08, and 0.937±0.11 for Aβ38, Aβ40 and Aβ42, respectively, with no significant differences between groups. The model predictions of fractional labeling of the Aβ peptides in the CSF are linearly related to the scaling factors, and thus sensitivity to this parameter in isolation is uninformative.
As will be shown below, this model is ‘system unidentifiable’. To reduce the number of parameters, the following assumptions were applied:
kCSF=kCSF38=kCSF40=kCSF42
kdel=kdel38=kdel40=kdel42
kret=kret38=kret40=kret42.
The first two parameters (kCSF and kdel) represent fluid flow processes and likely affect all three peptides equally. The third parameter (kret) could only be discerned for Aβ42 (see Appendix A) and may be different for Aβ38 and Aβ40. However, choosing the same value for kret for all three peptides allowed us to examine the extent of exchange of Aβ38 and Aβ40 relative to Aβ42. Exchange of Aβ38 and Aβ40 was found to be minimal and improved the fit of the model to the data in only a few subjects.
These assumptions reduced the model to 19 parameters. The CSF concentration size of each peptide is known, and because of steady state relationships Eqn. (11.5.4), Eqn. (11.5.5) and Eqn. (11.5.6), only 16 of the 19 parameters are independent. The choice of which three parameters are considered to be dependent is arbitrary, but the Aβ production rate constants kAb38, kAb40 and kAb42 are easily calculated (see Appendix E) and a convenient choice.
Most of the Aβ isotope-labeling curves were found to be well-fit by a simple model consisting of five ‘delay’ compartments arranged in series, with equal-valued rate constants for transfer between compartments, plus a single compartment turning over at a unique rate (see Appendix A). The data sets that could not be fit were primarily Aβ42 in subjects with significant amyloid plaque load as demonstrated by PET-PIB. The different morphology of the Aβ42 isotopic labeling time course compared to Aβ38 and Aβ40 in PIB− positive subjects is readily observed (e.g. see
Although the exact solution presented above incorporates known biology and physiology, the current dataset was unable to independently identify all 16 rate constants in the model, for reasons that will be clear following the system identifiability and sensitivity analysis below. Thus, the model was further simplified using the following assumptions:
Justifications for these assumptions are driven by the need to replace some of the poorly identified rate constants (i.e. kC99 and kAb) with kdel, thus producing a model that was quite similar to a simple five compartment delay that was known to be sufficient to fit the labeling curves in subjects without plaques (Appendix A). The irreversible loss rate constant of APP was poorly identified (vAPP) and its effects were lumped into vC99. It was further assumed that only half of C99 led to the production of Aβ38, Aβ40 and Aβ42. This is because Aβ peptides of other sizes are produced, with their abundance very roughly estimated from MALDI-TOF spectra of Aβ peptides in CSF [20]. It was further assumed that 50% of the irreversible loss of Aβ40 was to the CSF (i.e. v40=kCSF). Varying this fraction lost to the CSF between 10% and 90% had little effect on the results of the model (Appendix A). Finally, the return rate constant from the exchange compartment (kret) was set to 0.1 h−1. This was optimized using the three participants with the largest extent of Aβ42 exchange, using different fixed values of kret and determining which value gave the best fit to the labeling curves (Appendix A). The six imposed relationships reduced the total number of parameters from 19 to 14 (because kC99, vC99 and v40 were replaced by other parameters and vAPP and kret were set to specific values) and the number of independent parameters was reduced to 10 (an additional degree of freedom was lost by setting kAb38+kAb40+kAb42=½kdel). Choice of the four dependent parameters is arbitrary, but calculation of kAPP, kAb38, kAb40 and kAb42 from the other 10 parameters is illustrated in Appendix F. Exact solutions for the simplified model used in the previous publication are shown in Appendix G.
Although development of the simplified model was described in Example 11.6, the process was empirical. Using system identifiability analysis, a more rigorous approach is described here. The three transfer functions for the full model reveal that in principle 13 independent parameters may be determined from the labeling curve of each peptide (for methods, see references [11] and [12] and Appendix H). The full model has 25 parameters, demonstrating that the system is underdetermined. The assumptions from Example 11.5 of a common kCSF, kret and kdel for the three peptides were physiologically based and reduced the number of parameters to 19. The following parameters appear together as sums everywhere within the transfer functions: v38+kCSF, v40+kCSF, v42+kCSF, kC99+vAPP, and kAb38+kAb40+kAb42+vC99. This led to some of the assumptions of the simplified model, namely that kCSF is a constant fraction of v40, and that vAPP is zero. Additionally, if the SUM kAb38+kAb40+kAb42+vC99 is replaced with the one parameter kAb, the number of parameters is reduced to 14. Recognizing that kAPP does not appear in the rate equations for fractional labeling reduces the number of parameters to 13. Thus, these assumptions make the problem ‘system identifiable’ [11]. The 24 algebraic equations that appear in the transfer functions were not further manipulated to demonstrate ‘parameter identifiability’ due to their complexity. Rather, ‘practical identifiability’ issues with the model are demonstrated by the sensitivity analysis below, further motivating the reduction from 13 to 10 parameters in the simplified model.
Sensitivity analysis of the simplified model would not yield information about kC99 and kAb because these were explicitly replaced by kdel in the solution. Thus, the exact solutions to the full model were utilized in the sensitivity analysis. Sensitivity analysis was performed using parameters from a PIB− negative non-carrier and a PIB− positive presenilin-1 mutation carrier. Both participants were of similar age. The parameter values for the simplified model were originally optimized using the measured hourly plasma leucine enrichment data as the input. To simplify the sensitivity analysis, all of the parameters in the simplified model were re-optimized using the mathematical functions ƒ (Eqn. (11.2.1)) to describe plasma leucine values. The differences between the raw hourly plasma leucine data and the ƒ functions are shown in
The sensitivity analysis describes the sensitivity of the fractional labeling of Aβ42 in the third CSF compartment (pAβ42d3L) to changes in each of the major model parameters. For example, for kAb42, the sensitivity SkAb42 is:
This is obtained by taking the partial derivative of the exact solution for pAb42d3L Eqn. (11.3.8) with respect to kAb42. The sensitivity can be interpreted as:
for small ΔkAb42.
For the sensitivity analysis, the exact solutions become unbounded when kC99→kdel or kAb→kdel. To overcome this, the derivatives with respect to each of the parameters was taken and then the limit of the resulting equations was evaluated as kAb→kC99 and then kC99→kdel, applying L'Hôpital's rule when necessary. The detailed methods are described in Appendix J.
For the non-carrier, increases in v42 or kCSF also increased pAb42d3L between hours 5 and 24, with a maximum effect about 2 h prior to the peak enrichment of Aβ42. However, increases in v42 or kCSF decreased pAb42d3L between hours 24 and 36. The effects of increases in v42 or kCSF on actual kinetic curves are shown in
Returning to
The sensitivity of the FSR to parameter changes in the model parameters was also examined (
For both participants, the largest effect on ∂pAb42d3L/∂t (and thus FSR) came from v42 and kCSF. The next largest effect on FSR was from kex42, which had an opposite effect from v42 and kCSF. Thus, if both of these parameters are increased (as was noted in participants with plaques), they will tend to cancel each other out. The parameter kret had a modest effect on FSR, while the other parameters had even less effect.
The sensitivity of the monoexponential FCR was calculated (
In
The sensitivity of −∂ ln(p)/∂t to changes in parameters is presented in
The goal of the isotope-labeling study was to determine kAb42, which governs the production rate of Aβ42, and (v42+kCSF), which govern the irreversible loss rate of Aβ42. The sensitivity analysis demonstrated that most of the variation in the Aβ42 labeling curve between subjects is likely due to differences in v42, kCSF, kex42 and kret. However, kAb42 may be reliably estimated because it has a large and direct effect on the concentration of Aβ42 in CSF. The sensitivity of the CSF Aβ42 concentration is the derivative of Eqn. (11.5.6) with respect to the various parameters. For example, for kAb42:
The sensitivity of Aβ42 CSF concentration to changes in the different parameters is presented in Table 10. The most important parameters that evoke changes in the CSF concentration of Aβ42 are kAb42, v42 and kCSF. The production rate constant of Aβ42 from C99 (kAb42) was the most important parameter in determining the CSF concentration in the non-carrier, and second only to kCSF in the mutation carrier. Increases in kAb42 or kCSF are predicted to result in increases in the CSF concentration of Aβ42, whereas an increase in v42 causes a reduction in CSF Aβ42 concentration because v42 represents shunting of Aβ42 away from the CSF. For this reason, the model predicts that CSF Aβ42 concentration is decreased due to shunting to irreversible loss, perhaps including deposition into plaques. Thus, most of the information about the rate of production of Aβ42 is provided by the concentration of Aβ42 in CSF, while the shape of the isotopic enrichment curve tends to provide information about irreversible loss and exchange of Aβ42.
Sensitivity analysis is not helpful to analyze the effects of the scaling factors. However, the scaling factors affect the overall size of the fitted curve, which allows other parameters to be adjusted in combination to better fit different regions of the curve. Examining
The ratio of production rate constants for Aβ42 relative to Aβ40 is simply Eqn. (11.5.6) divided Eqn. (11.5.5):
This shows that if the CSF concentration ratio of Aβ42:Aβ40 is to remain constant, increases in irreversible loss of Aβ42 relative to Aβ40 must be accompanied by increases in production of Aβ42 relative to Aβ40. However, if production is held constant and irreversible loss of Aβ42 relative to Aβ40 increases, as may occur in the presence of plaques, then the CSF concentration of Aβ42 relative to Aβ40 will decline, as has been observed [13]. This equation also shows that an increase in production without an increase in irreversible loss (perhaps due to an absence of plaques) should result in an increase in CSF concentration of Aβ42 relative to Aβ40. This has also been observed in mutation carriers that are much younger than their expected age of onset [13]. An important observation is that exchange of Aβ42 has no impact on the steady state CSF concentration, because the flux of mass into the exchange compartment is identical to the flux of mass out if at a steady state.
The sensitivity analysis demonstrated that the overall shape of the Aβ labeling curves was affected by all of the parameters in the model, although some parameters had much larger effects than others. Previously, the FSR of the labeling curve between 5 and 14 hours was used to estimate production kinetics of Aβ peptides [7]. The sensitivity analysis demonstrates that the Aβ isotopic enrichment upslope is not highly affected by differences in production rate constants between subjects. Rather, the FSR likely reflected primarily irreversible loss and exchange, although no differences in FSR were found between Alzheimer's subjects and controls. However, in this region of the labeling curves, increased irreversible loss and increased exchange will tend to act in opposite directions, potentially canceling out each other's effects on FSR. On the other hand, as expected, the monoexponential FCR is strongly affected by the rate of Aβ irreversible loss in the absence of short-term exchange. However, the presence of exchange complicates the use of monoexponential FCR as a reliable measure of the true turnover rate. Much more information about the system is gleaned by fitting the entire time course to the new compartmental model, which is rooted in the biology and physiology of the system. In addition, the simultaneous use of CSF concentrations along with the labeling data allows determination of rate constants for both production and irreversible loss of Aβ peptides.
In other models, it was suggested that only the first 15 h of labeling data were required to fully describe the kinetics of the system [14]. While it is possible that the irreversible loss rate might be reasonably well estimated from the upslope of the labeling curve in normal participants, the current sensitivity analysis shows that the presence of exchange will affect the upslope of the curve, potentially muting the effects of increased irreversible loss (
The FSR has also been used to analyze the effects of α-secretase inhibitors on the labeling of Aβ peptides in humans and non-human primates [21] and [22]. Large changes in the upslope of the labeling curves were noted. This is not inconsistent with the present analysis. Although the sensitivity of the FSR to changes in production rate constants is small, it is not zero. In the case of inhibition of α-secretase, this should result in a large decrease in the production rate constants, resulting in a decrease in the FSR. As illustrated in
Several caveats about the compartmental model must be mentioned. Flow processes likely dictate the rate at which Aβ peptides transit from brain to the lumbar space. These processes are approximated here as a sequential series of compartments. More elaborate models that account for brain and subarachnoid space anatomy and flow may allow more accurate determination of the rates of Aβ peptide irreversible loss and production. Thus, the sensitivities reported here are those of the current compartmental model, not of the underlying biological system, which has yet to be fully elucidated. The current dataset is also not rich enough to identify the rates of production and irreversible loss of APP and C99. Additional kinetic data relevant to the production and irreversible loss of APP and C99 would certainly improve the estimation of Aβ production and irreversible loss rate constants. Also, measurement of concentrations of various Aβ peptides has a large impact on the estimates of the production rate constants for the Aβ peptides, and improvements in the precision and accuracy of concentration measurements would greatly aid future studies.
These data demonstrated that the FSR and monoexponential FCR previously used to characterize production and irreversible loss of Aβ peptides actually reflect the values of multiple parameters within a complicated system, and are not pure measures of production or irreversible loss. In steady-state studies, it is shown that estimation of the production rate is greatly enabled by combining isotope labeling data with concentration or pool sizes measurements. This also provides a mechanism for the observed decrease in CSF concentration of Aβ42 in Alzheimer's disease. The irreversible loss and exchange rate constants for Aβ peptides dominate the shape of the isotopic enrichment time course curve, and both constants may be readily determined by fitting the entire time course to the compartmental model. The later phases of the labeling process are better suited to resolve the irreversible loss and exchange processes of Aβ42. The conclusions of this study should enhance the design and interpretation of isotope-labeling experiments applied in the central nervous system.
Having described the invention in detail, it will be apparent that modifications and variations are possible without departing from the scope of the invention defined in the appended claims. Those of skill in the art should, however, in light of the present disclosure, appreciate that many changes could be made in the specific embodiments that are disclosed and still obtain a like or similar result without departing from the spirit and scope of the invention, therefore all matter set forth herein is to be interpreted as illustrative and not in a limiting sense.
While the present disclosure has been described with reference to various embodiments, it will be understood that these embodiments are illustrative and that the scope of the disclosure is not limited to them. Many variations, modifications, additions, and improvements are possible. More generally, embodiments in accordance with the present disclosure have been described in the context of particular implementations. Functionality may be separated or combined in blocks differently in various embodiments of the disclosure or described with different terminology. These and other variations, modifications, additions, and improvements may fall within the scope of the disclosure as defined in the claims that follow.
This application claims priority to PCT Patent Application PCT/US2013/071042 filed on Nov. 20, 2013, which claims priority to U.S. Provisional Patent Application No. 61/728,692 filed on Nov. 20, 2012, and entitled “METHODS OF DIAGNOSING AMYLOID PATHOLOGIES USING ANALYSIS OF AMYLOID-BETA ENRICHMENT KINETICS”, each of which is hereby incorporated herein by reference in its entirety.
This invention was made with government support under 5P01AG026276-S1 awarded by the National Institute on Aging, and R-01-NS065667 awarded by the National Institutes of Health. The government has certain rights in the invention.
Number | Date | Country | |
---|---|---|---|
61728692 | Nov 2012 | US |
Number | Date | Country | |
---|---|---|---|
Parent | PCT/US2013/071042 | Nov 2013 | US |
Child | 14717453 | US |