Patent Application
20210193322
The present invention relates to the application of network methods in investigating neurocognitive disorders.
Models of the human brain as a complex network of interconnected sub-units have improved the understanding of normal brain organization, and have made it possible to address functional changes in neurological disorders. These sub-units constitute so called brain modules, i.e. groups of regions that have a high density of connections within them, and with a lower density of connections between groups. It has been suggested that the modular organization of the brain underpins efficient integration between spatially segregated neural processes, which supports diverse cognitive and behavioural functions. Changes in brain networks can assist in identify patients with Alzheimer's disease (AD) and behavioural variant frontotemporal dementia (bvFTD).
For example, alterations in regional volumes have been identified in schizophrenia patients through study of structural networks in health and disease where pair-wise correlations of the cortical regional volume or thickness, as derived from in vivo measurements of T1-weighted magnetic resonance images (MRI), have been examined. This approach has shown clinical relevance by revealing alterations in regional volumes in schizophrenia patients. However, volumetric measures, which represent the product of cortical thickness (CT) and surface area (SA), may confound underlying differences. For example, consideration of changes in cortical thickness may provide insight into how disease alters the size, density, and arrangement of the cells within cortical layers. Changes in surface area, on the other hand, may provide information regarding disturbance in functional integration between groups of columns in diseased brains.
Previously, to monitor the effect of neuropharmacological interventions, whole brain or lobar volume has been used. However this is a relatively crude level of analysis.
As a further example of the use of networks in understanding the brain, as discussed in WO 2017/118733 (the entire contents of which is incorporated herein by reference), EEG data collected from a patient can be used to detect the intensity and directionality of electrical flow within the brain.
A poster entitled “Organization of cortical thickness networks in Alzheimer's disease and behavioural variant frontotemporal dementia across brain lobes” was presented by Vuksanovic et al. at the 6th Cambridge Neuroscience Symposium, Neural Networks in Health and Disease Sep. 7-8, 2017.
A further poster entitled “Divergent changes in structural correlation networks in Alzheimer's disease and behavioural variant frontotemporal dementia” was presented by Vuksanovic et al. at the ARUK Conference 2018, 20-21 March, London, UK.
A presentation entitled “Modular organization of cortical thickness and surface area structural correlation networks in Alzheimer's disease (AD) and behavioural variant frontotemporal dementia (bvFTD)” was given by Vuksanovic, V, at the 10th SINAPSE Annual Scientific Meeting 25 Jun. 2018 Edinburgh.
In a first aspect, the invention provides a method of determining patient response to a neuropharmacological intervention, comprising the steps of:
Optional features of the invention will now be set out. These are applicable singly or in any combination with any aspect of the invention.
By correlation matrix, it may be meant that a structural correlation network is generated which may then be represented by a matrix.
In one embodiment, the patient response may be in the context of a clinical trial e.g. for assessing the efficacy of a pharmaceutical in the treatment of a neurocognitive disease. Thus the patient group (plurality of patients) may be treatment group who have been diagnosed with the disease, or maybe a control (‘normal’) group. Ultimately the efficacy of the pharmaceutical may be assessed in whole or in part based on the patient group response determined in accordance with the present invention, optionally compared with a comparator group who have not received the intervention.
The physical structure measured or obtained may be cortical thickness and/or surface area. The values for the cortical thickness and/or surface area may be averaged values obtained from the structural neurological data. The structured neurological data may be acquired from magnetic resonance imaging (MRI) data or computed tomography data for each patient. The structural neurological data and the further structural neurological data are obtained at different points in time. As discussed herein, the structural neurological data may be obtained via magnetic resonance imaging, computed tomography, or positron emission tomography for each patient. These techniques are well known per se to those skilled in the art—see e.g. Mangrum, Wells, et al. Duke Review of MRI Principles: Case Review Series E-book. Elsevier Health Sciences, 2018, and “Standardized low-resolution electromagnetic tomography (sLORETA): technical details” Methods Find Exp. Clin. Pharmacol. 2002:24 Suppl. D:5-12; Pascual-Marqui R D etc.
The plurality of cortical regions may be at least 60, or at least 65. For example, 68. The cortical regions may, for example, be those provided by the Desikan-Killiany Atlas (Desikan et al. 2006).
A p-value may be determined for each pair-wise correlation across a plurality of subjects, and may be compared to a significance level, wherein only p-values less than the significance level are used to generate the corresponding correlation matrix. In determining the pair-wise correlation between pairs of structure nodes, the corresponding values of each structure node may be compared to a reference value and their co-variance determined. The significance level may be referred to as alpha (‘α’).
Comparing the first correlation matrix and the second correlation matrix may include comparing a number and/or density of inverse correlations in the first correlation matrix to a number and/or density of inverse correlations in the second correlation matrix. In making the comparison, groups of structure nodes corresponding to a same lobe may be identified, and the comparison made between the first and second correlation matrix may utilize the same lobe.
Assigning the plurality of structure nodes corresponding to cortical regions of the brain may further include defining groups which contain structure nodes corresponding to homologous or non-homologous lobes. Comparing the first correlation matrix and the second correlation matrix may include comparing the number and/or density of correlations between different groups of structure nodes. Said another way, comparing the first and second correlation matrices may include comparing correlations between pairs of structure nodes which are non-homologous.
In some examples, comparing the first correlation matrix and the second correlation matrix may include comparing the number and/or density of correlations between groups of structure nodes located respectively in the frontal lobe (anterior nodes) and the parietal and occipital lobes (posterior nodes). It has been found that, in examples of efficacious neuropharmacological intervention, the number and/or density of inverse correlations between anterior and posterior nodes decreases. As inverse correlations are postulated to indicate compensatory link formulation whereby atrophy in one node is associated with hypertrophy in a functionally linked node, it will be appreciated that a decrease in the number and/or density of inverse correlations indicates a decrease in the number of compensatory links.
Typically, the neurocognitive disease or cognitive disorder is a neurodegenerative disorder causing dementia, for example a tauopathy.
The patients may have been diagnosed with a neurocognitive disease, for example Alzheimer's disease or behavioural-variant frontotemporal dementia. The disease may be mild or moderate Alzheimer's disease. The disease may be a mild cognitive impairment. However the findings of the present inventors described herein have applicability to other neurocognitive diseases also.
Diagnosis criteria and treatment of tauopathies, and other neurocognitive diseases, are known in the art and discussed, for example, in WO2018/019823, and references cited therein.
The disease may be behavioural variant frontotemporal dementia (bvFTD). Diagnostic criteria and treatment of bvFTD is discussed, for example, in WO 2018/041739, and references cited therein.
As explained herein, the topology of the disturbance in structural network is different in these two disease conditions (AD and bvFTD) and both are different from normal aging. The changes from normal are global in character, and are not restricted to fronto-temporal and temporo-parietal lobes respectively in bvFTD and AD and indicate an increase in both global correlation strength and in particular non-homologous inter-lobar connectivity defined by inverse correlations.
These changes appear to be adaptive in character, reflecting coordinated increases in cortical thickness and surface area that compensate for corresponding impairment in functionally linked nodes. The effects were more pronounced in the cortical thickness network in bvFTD and in the surface area network in AD.
The inventors have observed that an important change distinguishing both forms of dementia from normal elderly controls is the emergence of significant inverse correlation networks linking anterior and posterior brain regions which may relate to functional adaptations or compensations for impairment due to pathology. Specifically, inverse correlations are postulated to indicate compensatory link formation whereby atrophy in one node is associated with hypertrophy in a functionally linked node, it will be appreciated that a decrease in the number and/or density of inverse correlations indicates a decrease in the number of compensatory links.
Thus, if a neuropharmacological intervention is efficacious, it is expected that the network organization will be brought back towards that observed with a normal (non-disease) comparator population. If the condition is treated at an early enough stage, the network organization may be brought back to something entirely equivalent to the normal control. The method therefore provides an objective means of distinguishing disease-modifying treatments from symptomatic treatments: symptomatic treatments accentuate the abnormal network architecture, and may indeed accentuate the risk of transmission of (for example) prion-like disease processes to healthy brain regions. Conversely, disease-modifying drugs work in the opposite direction, reducing the need for compensatory input from relatively less impaired brain regions by normalizing function in regions affected by pathology.
It will be appreciated, in the light of the disclosure herein, that analysis of structural or network organization has particular utility in providing more power in clinical trials, thereby allowing for the use of fewer subjects and shorter treatment times. In particular, in diseases such as mild AD, mild cognitive impairment and pre-mild cognitive impairment etc., the clinical trial end-points (cognitive and function) can be relatively insensitive and so require large numbers of subjects and\or longer time periods (see WO2009/060191).
Thus, typically, the neuropharmacological intervention will be a pharmaceutical intervention.
The neuropharmacological intervention may be a symptomatic treatment. Such compounds include acetylcholinesterase inhibitors (AChEIs)—these include tacrine, donepezil, rivastigmine, and galantamine. A further symptomatic treatment is memantine. These treatments are described in WO2018/041739.
As explained above, the inventors have found an increase in compensatory networks (number and/or density of non-homologous inverse correlations present in patient groups who have received such treatments).
The neuropharmacological intervention may be a disease modifying pharmaceutical rather than a symptomatic one. These treatments can be distinguished, for example, based on what happens when a patient is withdrawn from active treatment. Symptomatic agents defer the symptoms of the disease without affecting the fundamental disease process and do not change (or at least do not improve) the rate of longer term decline after an initial period of treatment. If, after withdrawal, the patient reverts to where they would have been without treatment, the treatment is deemed to be symptomatic (Cummings, J. L. (2006) Challenges to demonstrating disease-modifying effects in Alzheimer's disease clinical trials. Alzheimer's and Dementia, 2:263-271).
For example, a disease modifying treatment may be an inhibitor of pathological protein aggregation such as a 3,7-diaminophenothiazine (DAPTZ) compound. Such compounds are described in WO2018/041739, WO2007/110627, and WO2012/107706. The latter describes leuco-methylthioninium bis(hydromethanesulfonate) also known as leuco-methylthioninium mesylate (LMTM; USAN name: hydromethylthionine mesylate).
The contents of all of these WO publications in relation to the DAPTZ compounds they define are specifically included by cross reference.
Treatment with LMTM has been shown to reduce compensatory network correlations (especially non-homologous, positive and inverse correlations).
The neuropharmacological intervention may be a disease modifying pharmaceutical, and efficacy may be established by reduction in number and/or density of correlations between anterior and posterior brain regions of the first correlation matrix and the second correlation matrix.
Thus it can be concluded that in examples of efficacious neuropharmacological intervention (for example a disease modifying treatment) the number and/or density of inverse correlations between anterior and posterior nodes decreases.
The invention may also be utilized for identifying functional adaptations or compensations for impairment due to pathology in a patient population, for example to investigate “cognitive reserve”. The invention may be used in combination with conventional diagnostic or prognostic measures. These measures include the Alzheimer's Disease Assessment Scale-cognitive subscale (ADAS-Cog), National Institute of Neurological and Communicative Disorders and Stroke-Alzheimer's Disease and Related Disorders Association (NINCDS-ADRDA), Diagnostic and Statistical Manual of Mental Disorders, 4th Edn (DSMIV), and Clinical Dementia Rating (CDR) scale.
As explained above, the method of determining patient response to a neuropharmacological intervention may in turn be used to assess different patient cohorts in clinical trials of the neuropharmacological intervention. For example, the method may be for determining the effectiveness of a neuropharmacological intervention in a patient group. The method may be used for defining a patient group according to their patient response (e.g. in terms of the correlations/inverse correlations determined). The patient group may be identified in relation to their prior use of the neuropharmacological intervention, and optionally selected for further treatments appropriate to the patient response.
In a second aspect, the invention provides a method of determining a patient's likelihood of developing one or more neurological disorders, comprising the steps of:
The inventors have shown that even quite brief analysis using (for example) EEG of the brain can be used to potentially identify patients who are susceptible to one or more neurocognitive diseases (for example AD). Specifically, such individuals (patients or subjects, the terms are used interchangeably) may have a relatively high number of ‘sinks’, or sinks which are relatively strong, in the posterior lobes, and a relatively high number of ‘sources’, or sources which are relatively strong, in the temporal and/or frontal lobes. In apparently normal or prodromal subjects, in preferred embodiments, the method may be more sensitive than commonly used psychometric measures for determining such risk.
Optional features of the invention will now be set out. These are applicable singly or in any combination with any aspect of the invention.
The likelihood of a patient developing one or more neurological disorders may be referred to as a patient's susceptibility to one or more neurological disorders. The method may include a step of defining a state for each node, whereby a node is defined as either a sink or a source based on the calculated difference.
The network may be a renormalized partial directed coherence network. Any of the steps of the method may be performed offline, i.e. not live on a patient. For example, obtaining the data may be performed by receiving, over a network, data which has been previously recorded from a patient.
The data indicative of electrical activity within the brain may be electroencephalography data. The electroencephalography data may be β-band electroencephalography data. The data indicative of electrical activity within the brain may also be magnetoencephalography data or functional magnetic resonance imaging data.
Determining the patient's susceptibility may be performed using a machine learning classifier. For example, Markov models, support vector machines, random forest, or neural networks.
The method may include a step of producing a heat-map based, at least in part, on the states of the nodes, said heat-map indicating the location and/or intensity of nodes defined as sinks and nodes defined as sources within the brain of the patient. This representation of the defined nodes can aid (e.g. ergonomically) in the determination of the patient's susceptibility.
In determining the patient's susceptibility, a comparison may be made between the number and/or intensity of sources within the parietal and/or occipital lobes and the number and/or intensity of sinks within the frontal and/or temporal lobes. It has been seen experimentally that patients who are susceptible to one or more neurodegenerative diseases (and particularly Alzheimer's disease) have a relatively high intensity of sinks in the posterior lobes, and a relatively high intensity of sources in the temporal and/or frontal lobes.
The method may further comprise a step of deriving, using the states of the nodes, an indication of a degree of left-right asymmetry in the location and/or intensity of nodes in the brain corresponding to sinks and sources.
The neurological disorder may be a neurocognitive disease, which may be Alzheimer's disease.
The patient's susceptibility to one or more neurological disorders may be determined by comparing the number and/or intensity of nodes defined as sinks in the posterior lobe to a predetermined value, and/or comparing the number and/or intensity of nodes defined as sources in the temporal and/or frontal lobes to a predetermined value. A patient may be determined to be at a high risk of susceptibility if the number and/or intensity of nodes defined as sinks in the posterior lobe exceeds a predetermined value and/or if the number and/or intensity of nodes defined as sources in the temporal and/or frontal lobes exceeds a predetermined value (for example based on a ‘control’ subject or subjects established as having a low risk, or reference data (e.g. historical reference data) obtained from the same). Said another way, and more generally, a determination as to the susceptibility may be based on whether the patient has more and/or stronger sources and/or sinks in one region of the brain relative to another region of the brain. For example, if there are more and/or stronger sources in the temporal and/or frontal lobes than expected, and/or whether the patient has more and/or stronger sinks in the posterior lobe than expected based on data from control subjects, the patient may be determined as at risk of a neurodegenerative disorder. Such data from control subjects may have been established by longitudinal monitoring following base-line assessment.
The inventors have further observed that symptomatic treatment(s) increase activity outgoing from the frontal lobe compared to the non-medicated group.
The method of the invention according to this aspect may be used for assessing, testing, or classifying a subject's susceptibility to one or more neurological disorders for any purpose. For example, the score value or other output of the test may be used to classify the subject's mental state or disease state according to predefined criteria.
The subject may be any human subject. In one embodiment, the subject may be one suspected of suffering a neurocognitive disease or disorder e.g. a neurodegenerative or vascular disease as described herein, or maybe one who is not identified as at risk.
In one embodiment, the method is for the purpose of the early diagnosing or prognosing of a cognitive impairment, for example a neurocognitive disease, in the subject.
The disease may be mild to moderate Alzheimer's disease.
The disease may be mild cognitive impairment.
However the findings of the present inventors described herein have applicability to other neurocognitive diseases also. For example the disease may be a different dementia, for example vascular dementia.
The method may optionally be used to inform further diagnostic steps or interventions for the subject—for example based on other methods of imaging or invasive or non-invasive biomarker assessments, where such methods are known per se in the art.
In some embodiments, the method may be for the purpose of determining the risk of a neurocognitive disorder in the subject. Optionally, said risk may additionally be calculated using further factors, e.g. age, lifestyle factors, and other measured physical or mental criteria. Said risk may be a classification of “high” or “low” or may be presented as a scale or spectrum.
It will be apparent from the disclosure herein that in addition to assessing the likelihood of developing one or more neurological disorders, the same methodology can be used to assess the efficacy of a disease-modifying treatment to reduce said risk and/or treat said disease i.e. to assess the efficacy of a pharmaceutical for prophylaxis or treatment of the disease or disorder. This may optionally be in the context of a clinical trial as described herein, e.g. in comparison to a placebo, or other normal control.
Specifically, the disclosure herein indicates that the methods of the invention (e.g. based on EEG technology) can provide a powerful and sensitive measures of the disease impact on a subject. This opens up the opportunity to demonstrate the efficacy of a disease-modifying treatment in smaller groups of subjects (e.g. less than or equal to 200, 150, 100, or 50 in treatment and comparator arms) and over a shorter interval (e.g. less than or equal to 6, 5, 4, or 3 months) and in earlier stages of disease or less severe disease (e.g. prodromal AD, MCI or even pre-MCI) than is possible using currently available methods.
Thus, as discussed above, the method may be used with different patient cohorts in clinical trials of a neuropharmacological intervention e.g. a patient group (plurality of patients) may be a treatment group who have been diagnosed with the disease (for example early stage disease) treated with a putative disease modifying treatment vs. group treated with placebo.
Thus in one further aspect the method steps of the second aspect are used to determine disease status or severity in a patient, rather than determining a patient's likelihood of developing one or more neurological disorders. That status can in turn be monitored as part of clinical management or a clinical trial.
Thus one further aspect of the invention there is provided a method of determining a patient response to a neuropharmacological intervention against a neurological disorder, comprising the steps, before the neuropharmacological intervention, of:
Thus these methods of the second and further aspects (and corresponding systems discussed hereinafter) can be used for both clinical trials and clinical management. In terms of clinical management, a high degree of certainty (for example, 70%, 80%, 90%, or 95% probability) that the electrical activity within the brain of the patient (e.g. as assessed using EEG) is abnormal in a “normal” person (i.e. presently undiagnosed) may be a strong indication for immediately starting dementia medication treatment. The EEG could also be used in at intervals, for example, 1, 2, 3, 4, 5, or 6 months to monitor response to treatment. Conversely, a person with lower probability of abnormal EEG (for example, 30%, 40%, 50%, 55%, or 60%) could be followed more closely at monthly, bimonthly, or trimonthly intervals. Further tests by other means appropriate to the disorder, such as are known in the art (e.g. assessment of biomarkers based on amyloid or tau PET or CSF) may optionally be used in conjunction with the method.
Optional features in relation to the methods of the second aspect apply mutatis mutandis to this aspect.
In a third aspect, the invention provides a system for determining patient response to a neuropharmacological intervention, the system comprising:
Optional features of the invention will now be set out. These are applicable singly or in any combination with any aspect of the invention.
By correlation matrix, it may be meant that a structural correlation network is generated which may then be represented by a matrix.
The physical structure measured or obtained may be cortical thickness and/or surface area. The values for the cortical thickness and/or surface area may be averaged values obtained from the structural neurological data. The structured neurological data may be acquired from magnetic resonance imaging (MRI) data or computed tomography data for each patient. The structural neurological data and the further structural neurological data are obtained at different points in time. As discussed herein, the structural neurological data may be obtained via magnetic resonance imaging, computed tomography, or positron emission tomography for each patient. These techniques are well known per se to those skilled in the art—see e.g. Mangrum, Wells, et al. Duke Review of MRI Principles: Case Review Series E-book. Elsevier Health Sciences, 2018, and “Standardized low-resolution electromagnetic tomography (sLORETA): technical details” Methods Find Exp. Clin. Pharmacol. 2002:24 Suppl. D:5-12; Pascual-Marqui R D etc.
The plurality of cortical regions may be at least 60, or at least 65. For example, 68. The cortical regions may, for example, be those provided by the Desikan-Killiany Atlas (Desikan et al. 2006).
The display means may provide each of the first correlation matrix and the second correlation matrix on a display, wherein correlation values in each correlation matrix are given a colour corresponding to the relative amplitude or strength of the correlation.
The verification means may be configured to determine a p-value for each pair-wise correlation, and compare the p-value for each pair-wise correlation, and may compare the p-value to a significance level, the correlation matrix generating means may be configured to use only p-values less than the corrected significance level when generating a correlation matrix. The significance level may be referred to as alpha (‘α’).
The comparison means may be configured to compare a number and/or density of inverse correlations in the first correlation matrix to a number and/or density of inverse correlations in the second correlation matrix. In making the comparison, groups of structure nodes corresponding to a same lobe may be identified, and the comparison made between the first and second correlation matrix may utilize the same lobe.
Assigning the plurality of structure nodes corresponding to cortical regions of the brain may further include defining groups which contain structure nodes corresponding to homologous or non-homologous lobes. The comparison means may be configured to compare the first correlation matrix and the second correlation matrix by comparing the number and/or density of correlations between different groups of structure nodes. Said another way, comparing the first and second correlation matrices may include comparing pairs of structure nodes which are non-homologous.
The comparison means may be configured to compare the first correlation matrix and the second correlation matrix by comparing the number and/or density of correlations between groups of structure nodes located respectively in the frontal lobe (anterior nodes) and the parietal and occipital lobes (posterior nodes). It has been found that, in examples of efficacious neuropharmacological intervention, the number and/or density of inverse correlations between anterior and posterior nodes decreases. As inverse correlations are postulated to indicate compensatory link formulation whereby atrophy in one node is associated with hypertrophy in a functionally linked node, it will be appreciated that a decrease in the number and/or density of inverse correlations indicates a decrease in the number of compensatory links.
In one embodiment, the patient response may be in the context of a clinical trial e.g. for assessing the efficacy of a pharmaceutical of a neurocognitive disease. Thus the patient group (plurality of patients) may be treatment group who have been diagnosed with the disease, or maybe a control (‘normal’) group. Ultimately the efficacy of the pharmaceutical may be assessed in whole or in part based on the patient group response determined in accordance with the present invention.
As explained in relation to the first aspect, the neurocognitive disease will generally be a neurodegenerative disorder causing dementia, for example a tauopathy.
The patients may have been diagnosed with the neurocognitive disease, for example Alzheimer's disease or behavioural-variant frontotemporal dementia. The disease may be mild or moderate Alzheimer's disease. The disease may be a mild cognitive impairment
Diagnosis criteria and treatment of tauopathies, and these disorders, are discussed, for example, in WO2018/019823, and references cited therein.
The disease may be behavioural variant frontotemporal dementia (bvFTD). Diagnosis criteria and treatment of bvFTD is discussed, for example, in WO 2018/041739, and references cited therein.
As explained herein, the topology of the disturbance in structural network is different in the two disease conditions (AD and bvFTD) and both are different from normal aging. These changes appear to be adaptive in character, reflecting coordinated increases in cortical thickness and surface area that compensate for corresponding impairment in functionally linked nodes.
Thus, if a neuropharmacological intervention is efficacious, it is expected that the network organization will be brought back towards a normal state. If the condition is treated at an early enough stage, the network organization may be brought back to normal indicating arrest or reversal of the disease state. The system therefore provides an objective means of distinguishing disease-modifying treatments from symptomatic treatments as described above.
Typically, the neuropharmacological intervention will be a pharmaceutical intervention.
The neuropharmacological intervention may be a symptomatic treatment as described above.
For example, a disease modifying treatment may be an inhibitor of pathological protein aggregation such as a 3,7-diaminophenothiazine (DAPTZ) compound as described above.
In a fourth aspect, the invention provides a system for determining a patient's susceptibility to one or more neurological disorders, the system comprising:
As described above in relation to the second aspect, the inventors have shown that even quite brief analysis using (for example) EEG of the brain can be used to potentially identify patients who are susceptible to one or more neurocognitive diseases (for example AD).
The system can be used for both clinical trials and clinical management.
Thus in a further aspect there is provided a system as described above for determining a patient response to a neuropharmacological intervention against a neurological disorder, In this aspect the determination means system may be configured for determining, using the calculated differences, the patient's status in relation to the neurological disorder.
The system can be used to determine a further status of the patient after the neuropharmacological intervention, and optionally configured to determine, based on the first and one or more subsequent statuss, the patient response to the neuropharmacological intervention, as described above in relation to the corresponding method.
Other optional features of the invention will now be set out. These are applicable singly or in any combination with any aspect of the invention.
The system may include state definition means configured to define a node as either a sink or a source based on the calculated difference.
The network may be a renormalized partial directed coherence network. The system may operate “offline”, i.e. not live on a patient. For example, obtaining the data may be performed by receiving, over a network, data which has been previously recorded from a patient.
The data indicative of electrical activity within the brain may be electroencephalography data. The electroencephalography data may be β-band electroencephalography data. The data indicative of electrical activity within the brain may also be magnetoencephalography data or functional magnetic resonance imaging data.
The determination means may be configured to use a machine learning classifier to determine the patient's susceptibility to one or more neurological disorders. For example, Markov models, support vector machines, random forest, or neural networks.
The display means may be configured to present a heat map indicative of the location and/or intensity of nodes defined as sinks and nodes defined as sources within the brain. This representation of the defined nodes can aid (e.g. ergonomically) in the determination of the patient's susceptibility.
The system may further comprise a heat map generating means, configured to produce a heat map based at least in part on the states of the nodes, said heat-map indicating the location and/or intensity of nodes defined as sinks and nodes defined as sources within the brain of the patient. This representation of the defined nodes can aid (e.g. ergonomically) in the determination of the patient's susceptibility.
The determination means may compare the number/and or intensity of sources within the parietal and/or occipital lobes as compared to the number and/or intensity of sinks within the frontal and/or temporal lobes. It has been seen experimentally that patients who are susceptible to one or more neurodegenerative diseases (and particularly Alzheimer's disease) have a relatively high number and/or intensity of sinks in the posterior lobes, and a relatively high number and/or intensity of sources in the temporal and/or frontal lobes.
The system may further comprise an asymmetry map generation means, configured to derive, using the states of the nodes, an indication of a degree of left-right asymmetry in the location and/or density of nodes in the brain corresponding to sinks and sources.
The neurological disorder may be a neurocognitive disease, which is optionally Alzheimer's disease.
The determination means may compare the number and/or intensity of nodes defined as sinks in the posterior lobe to a predetermined value, and/or compare the number and/or intensity of nodes defined as sources in the temporal and/or frontal lobes to a predetermined value. The determination means may determine a patient to be at a high risk of susceptibility if the number and/or intensity of nodes defined as sinks in the posterior lobe exceeds a predetermined value and/or if the number and/or intensity of nodes defined as sources in the temporal and/or frontal lobes exceeds a predetermined value. Said another way, and more generally, a determination as to the susceptibility may be based on whether the patient has more and/or stronger sources and/or sinks in one region of the brain relative to another region of the brain. For example, if there are more/and/or stronger sources in the temporal and/or frontal lobes than expected, and/or whether the patient has more and/or stronger sinks in the posterior lobe than expected, then the patient may be determined as at risk of a neurodegenerative disorder.
The inventors have further observed that symptomatic treatment(s) increase activity outgoing from the frontal lobe compared to the non-medicated group.
The system of the invention according to this aspect may be used for assessing, testing, or classifying a subject's susceptibility to one or more neurological disorders for any purpose. For example, the score value or other output of the test may be used to classify the subject's mental state or disease state according to predefined criteria.
The subject may be any human subject. In one embodiment, the subject may be one suspected of suffering a neurocognitive disease or disorder e.g. a neurodegenerative or vascular disease as described herein, or maybe one who is not identified as at risk.
In one embodiment, the system is for the early diagnosing or prognosing of a cognitive impairment, for example a neurocognitive disease, in the subject as described above.
The system may optionally be used to inform further diagnostic steps or interventions for the subject—for example based on other systems for imaging or invasive or non-invasive biomarker assessments, where such systems are known per se in the art.
In some embodiments, the system may be for determining the risk of a neurocognitive disorder in the subject. Optionally, said risk may additional be calculated using further factors, e.g. age, lifestyle factors, and other measured physical or mental criteria. Said risk may be a classification of “high” or “low” or may be presented as a scale or spectrum.
As with the methods described herein, the system may be used in the context of a clinical trial, to assess the efficacy of a neuropharmacological intervention. The system may be used to demonstrate the efficacy of a disease-modifying treatment, for example LMTM, in a relatively small number of subjects (e.g. 50) over a relatively short time scale (e.g. 6 months) and in early disease stages (for example mild cognitive impairment or possible pre-mild cognitive impairment).
Further aspects of the present invention provide: a computer program comprising executable code which, when run on a computer, causes the computer to perform the method of the first or second aspect; a computer readable medium storing a computer program comprising code which, when run on a computer, causes the computer to perform the method of the first or second aspect; and a computer system programmed to perform the method of the first or second aspect. For example, a computer system can be provided, the system including: one or more processors configured to: perform the method of the first or second aspect. The system thus corresponds to the method of the first or second aspect. The system may further include: a computer-readable medium or media operatively connected to the processors, the medium or media storing computer executable instructions corresponding to the method of the first or second aspects.
Embodiments of the invention will now be described by way of example with reference to the accompanying drawings in which:
Aspects and embodiments of the present invention will now be discussed with reference to the accompanying figures. Further aspects and embodiments will be apparent to those skilled in the art. All documents mentioned in this text are incorporated herein by reference
The subjects discussed in this document participated in three global Phase 3 clinical trials which have now been completed. Two of the clinical trials were in mild to moderate AD (Gauthier et al., 2016; Wilcock et al., 2018) and the third was from a large study of bvFTD (Feldman et al., 2016). Comparable data was available from well characterized healthy elderly (HE) subjects participating in an ongoing longitudinal study of the Aberdeen 1936 Birth Cohort (ABC36) (Murray et al., 2011). In all, in the examples discussed herein, there were 628 subjects with 213 in each of the dementia groups and 202 healthy elderly subjects. The bvFTD patients were diagnosed according to the International Consensus Criteria for bvFTD, with mild severity on the Mini-Mental State Examination (MMSE) score of 20-30 inclusive. AD patients were diagnosed according to the criteria from the National Institute of Aging and the Alzheimer's Association, with mild to moderate severity defined by an MMSE score of 14-26 (inclusive) and a Clinical Dementia Rating (CDR) total score of 1 or 2. They were drawn from the corresponding larger group (N=1132) to match the number of participants in the bvFTD group. The healthy elderly (HE) subjects were selected from a well characterized Aberdeen 1936 Birth Cohort.
The multi-side source imaging data sets, used to generate the correlation matrices discussed below, were standard T1-weighted MRI images acquired using equivalent manufacturer specific 3DT1 sequences. The data from trial patients were pooled to permit overall group-wise comparisons. The train scanners were limited to 1.5T and 3T (30%) field strengths from three manufacturers (Philips, GE, and Siemens). MRI images in the ABC36 cohort were all acquired using the same (Philips) 3T scanner. The images were processed using an automated process pipeline implemented in a manner known per se. In addition to the volume-based methods of image processing, the pipeline produces surface-based regional measurements of cortical morphology such as thickness, the local curvature or surface area. An example of an automated processing pipelines suitable for the above methods is FreeSurfer v5.3.0 available from the Athinoula A. Martinos Centre for Biomedical Imaging at Massachusetts General Hospital.
Surface area was calculated from the imaging data sets using triangular tessellation of the grey/white matter interface and white matter/cerebrospinal fluid boundary (referred to as the pial surface). The cortical thickness was calculated as an average of the distance from the white matter surface of the closest point on the pial surface, and from that point back to the closest point to the white matter surface. A parcellation scheme, known per se, was used to extract cortical thickness and surface area of 68 cortical regions from both hemispheres based on the Desikan-Killiany Atlas. A list of regions and their lobar assignment is given in Table A.1 in Annex A.
In these figures, zero entries correspond to non-significant correlations. It was found that the significant network correlations had both positive and negative values (see
The networks, represented in these figures as correlation matrices, can be constructed by correlating either surface area or cortical thickness across all subjects within a particular diagnostic category (i.e. HE, bvFTD, and AD). A cortical region (as defined by the Desikan-Killiany brain Atlas) represents a node and a pair-wise correlation between nodes represents a graph edge or link/connection was constructed correlating either SA or CT across all participants within each diagnostic category. Each correlation matrix was calculated based on S×N array containing N regional CT/SA values from S subjects within each group. In this way, six N×N (e.g. 68×68) correlation matrices were obtained (one CT or SA structural correlation matrix for each study group). The matrix element eij is the value of the partial correlation between the region i and j (i, j=1, 2, . . . N) (i.e. between vectors xi and xj that contain regional measurements from subjects within each group). The partial correlations were calculated as linear, Pearson's correlation coefficients between pairs of xi and xj after first removing the effects of all other regions m≠(i; j) and then adjusting both xi and xj for controlling variables (stored in a separate array S×C, where C represents the number of controlling variables). This means that prior to correlation analysis a linear regression was performed on every xi to remove the effect of age, gender, and mean CT (mean cortical thickness of all areas) or total surface area (sum of overall surface areas). Self-correlations (represented by the main matrix diagonnetwork measures were calculated on the lower triangular part of the matrix. The partial correlation, eij (i.e. edge weight), can be calculated according to the following general equations:
e
ij=ρi≡corr(xi, xj|xc)
Where xi;j denotes an array of variables and xc denotes any subset of conditioning variables. To arrive at this general form of the partial correlation, the process begins from i, j, c=1, 2, 3:
Hence, for any subset of c of conditioning variables:
In some examples, to verify that the network retained only statistically significant correlations, the calculated correlation coefficients were adjusted for multiple tests using the False Discovery Rate (FDR) procedure as set out in Storey, 2002. The FDR procedure tests each calculated p-value (from the pair-wise correlation calculation) against a corrected significance level, in this example α=0.05, and accepts only p-values smaller than the adjusted significance level as truly significant. Those pair-wise correlations that did not pass the FDR test may be set to zero; otherwise, all non-zero correlations, whether positive or negative, were retained (see
In this way, a 68×68 correlation matrix can be constructed for either CT or SA in each clinical group, which represents the structural correlation network for either surface area of cortical thickness. A matrix element quantities the strength of the correlation between cortical regions for either cortical thickness or surface area and it does not in itself represent an actual physical connection. In the context of structural correlation network analysis in neurodegenerative disorders, such correlations are considered to imply either a co-atrophy relationship (if positive) or an inverse atrophy/hypertrophy relationship (if negative) between brain regions.
With regards to the structural correlation networks and/or matrices generated using the above methods, it is useful to use the following measures to compare the structural network properties of the three clinical groups: edge strength, node degree, node within-module degree z-score, and participation index. Edge strength and node degree represent two basic networks attributes; they respectively quantify the correlation strength between nodes and the number of pairwise correlations for each node. To assess whether cortical lobes represent modules, two network measures were utilized which assess modularity in network interactions, namely within-module degree z-score and participation index. All measures (except node degree) were computed on weighted graphs and where estimated as averages across the four lobes (described below). The measures were computed on either binary or weighted graphs (as discussed below). From purely theoretical studies, it is known that the calculated topological properties of a network depend on the choice of the threshold value (van Wijk et al. 2010). In this document, a fixed threshold for each group-based correlation matrix is chosen.
Node Degree
Node degree, ki, represents the number of significant correlations for each node in the network. In general, node degree is calculated from a binarised correlation matrix where each significant correlation in the matrix is replaced with either 1 if it is significant or with 0 if it is not. Examples of binarised matrices are shown in
The degree of a node i, i.e. the number of significant links connected to a node, can be calculated as:
where N is the number of nodes, and aij represents the connection between nodes i and j having a value of 1 if there is a direct connection between nodes and 0 otherwise.
Modularity Index
Node participation index and within-module degree z-score assess the role of a node according to modules. Network modules (also known as community structures) represent densely connected sub-graphs of a network, i.e. subsets of nodes within which network connections are denser, and between which connections are sparser. It is useful to examine the modular organization of frontal, temporal, parietal, and occipital divisions of cortical thickness or surface area network as defined as modules. Since these lobar divisions of the cortical surface area are not necessarily modular in themselves, it may be necessary to first test whether lobar divisions are intrinsically modular. In one example, this may be done by calculating the modularity index (Q) of the networks according to each lobe. The modularity index quantifies the observed fraction of within- module degree values relative to those expected if connections were randomly distributed across the network. Since the constructed cortical thickness and surface area networks contain both positive and negative edge strengths, it is possible to use an asymmetric generalization of the modularity quality function. For example, as introduced in Rubinov and Sporns (2011):
Where ωij+ is equal to the i, j-th element of the correlation matrix, i.e. the strength of the pair-wise correlation between cortical regions, ωij if ωij>0 and is equal to zero otherwise.
Similarly, ω
It was found that lobar organization of the cortical surface into frontal, parietal, temporal, and occipital divisions is in fact modular (see Annex A). Accordingly, it is then possible to calculate the contribution of individual nodes to lobar modules as the node participation index and the within-modules z-score, which is referred to as node between-lobes participation index and node within-lobe z-score.
Node Between-Lobes Participation Index
In general, the participation index p assesses inter-modular connectivity. It may be considered the ratio of within-lobe node edges to all other lobar modules in the network, where node pi tends to 0 if the node has links exclusively within its own module, and tends to 1 if the node links exclusively outside of its own module. The weighted network participation is calculated by:
where M is the set of modules and kiw(m) is the weighted number of links of the i-th node to all other nodes in module m—inter-modular degree and kiw is the total degree of the i-th node. Within this document, the term between-lobes participation is used for this network measure.
Node Within-Lobe Degree z-Score
The complement of the between lobes participation index is the normalized within-lobe degree, zi, which assesses intra-lobar connectivity by means of z-score i.e. by the normalized deviation of the inter-lobar degree of a node with the respective mean degree distribution. Therefore, node within-lobe z-score, zi, is large for a node with more intra-modular connections relative to the inter-modular mean connectivity. For networks in which correlation strengths are preserved, the node within-module degree z-score is calculated as:
where kiw(m) is as above, ki−w(mi) is the mean of the within module mi degree distribution, and σk
Node Role in Network Modular Organization
Node role with the modular lobar organization depends on its position in the z-pi parameter space. There are four possible roles that anode can have in the network, which are assigned on the basis of higher than average measures of nodal properties. It is useful to consider two of these roles, so called connectors or global network hubs (that have high between-lobes participation and high within-lobe degree z-score) and so called provincial hubs (that have high within-lobe degree z-score and low between-lobes participation). The thresholds for high and low values of zi and pi were set above 1.5 and 0.05, respectively.
Statistical Analysis
Statistical differences in demographic and cognitive scores in subjects were assessed using either one-way analysis of variance or two-tailed t-tests. The data was checked for normality of distribution using a one-sample Kolmogorov-Smirnov test. A chi-square test was used to test for differences in distribution of males and females between the groups. Statistical differences in global network correlation strength according to diagnostic groups were tested using one-way analysis of variance for unbalanced sample size (to account for an uneven number of significant correlations across the networks). The node degree, within-lobe z-score, zi, and between-lobes participation index, pi, were compared across the groups using the Kruskal-Wallis test, a non-parametric one-way analysis of variance test. Results were reported as significant at the level p<0.05.
Results
Table 1, below, shows demographic, cognitive, and mean CT and SA for each group according to clinical diagnosis. The 3 groups different significantly by age, AD patients being older than HE and bvFTD (p<10−4 in all tests). Significant differences were also seen in cognitive scores on the MMSE scale, AD patients being the most impaired and bvFTD more impaired than HE subjects (p<10−4 in all tests). The mean CT and total SA differed across groups.
The differences between HE and both patient groups were significant in terms of both mean CT (p<10−4, in both tests) and total SA (p<0.003 in both tests), but bvFTD and AD groups did not differ from each other. The mean CT and total SA values averaged by brain lobes are given in Table A.3 in Annex A. Therefore, although AD and bvFTD differ in terms of lobar distribution of pathology, age and severity of cognitive impairment, neither overall extent of cortical thinning or change in mean surface area provide a means of distinguishing between the two conditions.
Lobar Properties of Structural Correlation Network
As the definition of correlation-based network organization depends on the choice of the threshold value, it is useful to ensure that the networks defined herein were non-random in their global topology by calculating the density/sparsity value (κ). Brain networks are considered to show non-random (small-world) topology if κ>0.1, which was the case for all networks considered here. It is also useful to ensure that inverse correlations were not omitted after thresholding (see
Using the modularity index, an investigation was performed to determine whether cortical lobes as conventionally defined correspond to network modules in the CT network. It was found that only two homologue pairs (posterior cingulate and precentral cortex) in the CT network and two homologue pairs in the SA network (posterior cingulate cortex and paracentral and right banks of the superior temporal sulcus were miss-assigned in the modularity index algorithm. Table A.2 (in Annex A) gives details of the algorithm input and output. In practice, it is accepted that a Q value of above 0.3 is a good indicator of the existence of significant modules in a network. To estimate the confidence interval of the Q values for the data set, repeated calculations against 100 CT matrices generated on surrogate datasets was performed. Each of the 100 surrogate CT and SA matrices were generated by randomly drawing 213 subjects from the three study cohorts and calculating Q values on the correlation matrix obtained for CT and SA. The values of Q are shown in
It can therefore be concluded that the cortical lobes as conventionally described correspond to non-random modules in the CT network.
Mean Correlation Strength of the CT and SA Networks
The mean strength of networks of inverse correlations in the CT network also differed in frontal and temporal lobes, see
Nodal Measures in the CT Network
Node Degree
Node degree, which quantifies the mean number of significant positive correlations per node is shown averaged over frontal, temporal, parietal, and occipital lobes for the CT network in
There were significant differences between groups in frontal, temporal, parietal, and occipital lobes (p≤10−4 in all tests). Both bvFTD and AD subjects had higher node degree in frontal and temporal lobes (p<0.006 for all tests) compared with HE subjects. The bvFTD group had notably higher node degree in parietal and occipital lobes than the AD group (p≤0.02 for all tests). A similar pattern was found for the number of inverse correlations in the CT network in frontal, temporal, parietal, and occipital lobes (p≤0.02 in all tests). These differences were driven by a larger number of significant inverse correlations in bvFTD and AD than in the HE group across all four lobes (p<0.01 in all test). None of the differences between bvFTD and AD groups was significant.
Node Between-Lobes Participation Index
Group differences were found in the node between-lobes participation index for CT. The index measures the extent of significant positive correlation with nodes in different lobes. This was significant for lobes located in the temporal, parietal, and occipital lobes (p≤0.03 for all tests). The differences reflect higher index values relative to the HE group in the parietal (p<0.003 in both groups), temporal p=0.01 in AD) and occipital (p=0.002 in bvFTD) lobes. This is shown in
Nodal Measures in the SA Network
Node Degree
Node degree values in the SA network are shown for frontal, temporal, parietal, and occipital lobes in
Positive correlations differed between diagnostic groups in frontal, temporal, parietal, and occipital lobes (p≤0.03). As with the CT network, both bvFTD and AD groups had higher SA node degree than the HE group in frontal, temporal, and parietal lobes (p<10−4 in all tests). For the occipital lobe the only difference which was significant was between the AD and HE groups (p=0.04). In contrast to the CT network, the node degree in the parietal lobe was also significantly higher in AD than in bvFTD (p=0.004).
The inverse correlation SA network also showed significant group differences in frontal, temporal, parietal, and occipital lobes (p≤0.001 in all tests). Again, both bvFTD and AD groups had higher node degrees than the HE group in all four lobes (p<0.001 for all tests). In contrast to the CT inverse correlation network, the AD group had higher node degree than the bvFTD group in the frontal (p=0.02) and parietal (p=0.01) lobes.
Node Between-Lobes Participation Index
Both bvFTD and AD groups had higher index values than the HE group for the positive SA correlation network in all four lobes (p<10−4). In contrast to the CT correlation network, the inverse SA correlation network also showed significant differences in frontal and parietal lobes (p≤0.04 for both patient groups) and in temporal lobe for the AD group (p<0.001) relative to the HE group.
Hubs of the Structural Correlation Networks
CT Network Hubs
There are four possible combinations of mean values of between-lobes participation index (p high/low) and with-lobe z-score (high/low). Here, only the case of high between-lobes index and high within-lobe z-score are considered in order to focus on nodes high hub-like characteristics. Tables A.4-A.6 (see Annex A) provide data for the global and provincial network hubs. The remaining two combinations were examined, but were uninformative. The number and distribution of network hubs within high p and high z values in the positive CT correlation network differed between study groups. In the HE subjects, hubs were distributed across the whole cortex; each lobe had at least one hub, with four hubs in the frontal lobe. The re-organization of hub topology occurred differently in the two disease groups. This is shown in
Nodes with hub-like properties in the inverse correlation CT matrix were present exclusively in frontal and temporal lobes in all three groups and their topological distribution differed between the groups. See
SA Network Hubs
Hubs in the positive correlation SA network are shown in
Hubs in the inverse correlation SA network were present in either frontal or temporal lobe only in all three groups. However, the HE group had one hub in the parietal (precuneus) and bvFTD had two (inferiorparietal and paracentral) (see Table A.5 in Annex A). Interestingly, most of the inverse correlation SA hubs in AD were found in the frontal lobe.
Cortical Thickness—Cortical Surface Area Coupling Topology
The coupling strength between CT and SA nodes was calculated by element-wise multiplication of corresponding CT and SA correlation matrices.
Discussion
Baseline structural correlation networks in subjects diagnosed clinically with either bvFTD or AD in three large global clinical trials have been examined, and compared with healthy elderly subjects in a well-characterized birth cohort. For each group, networks were constructed from the partial correlations between 68×68 pairs of cortical surface regions (nodes) in terms of their thickness and surface area. The approach adopted has permitted a systematic analysis if both positive and inverse network correlations in the three clinical contexts. The methods and data discussed herein represent the first systematic comparative analysis of cortical thickness and surface area in a large population of subjects. Since the numbers needed to be comparable in the three groups, the overall study size was determined by the number of bvFTD subjects available. As this is a rare disease, it was necessary for the bvFTD component of the study to be global, with patients entering from 70 trial sites in 13 countries. With 213 patients included in the study, this represents the largest set of MRI scan data in bvFTD subjects available thus far. In order to match this, 213 patients were drawn randomly from a much larger group of 1131 AD patients coming from 116 sites in 12 countries for study TRx-237-005 and 128 sites in 16 countries for study TRx-237-015 (accessible for example from the US National Library of Medicine). The 202 normal elderly subjects come from a well-characterized birth cohort that has been studied longitudinally. The findings reported are therefore robust and can be considered representative of international populations meeting accepted diagnostic criteria.
Modularity of Networks By Lobes
It has been shown that the structural correlations in frontal, temporal, parietal, and occipital divisions of the cortical surface are inherently modular for both the cortical thickness and surface area networks. That is, the results confirm that the standard lobar divisions of the cortex share common network modularity attributes, such that they differ from what would be expected in a comparable random network. Modules of highly clustered networks confer so called ‘small-world’ network properties and are thought to provide an optimal balance between local specialization and global integration. The results from healthy elderly subjects are comparable with prior work in a smaller and younger healthy group revealing an underlying modular architecture in the regional thickness correlation network. The results also indicate that intrinsic lobe-wise modularity persists in both bvFTD and AD, indicating that the overall lobar architecture of the networks is preserved in the presence of neurodegenerative change. As discussed further below, this contrasts with the hub-like organization of the networks which changes in a disease-specific manner.
Similarities and Differences Between AD and bvFTD Relative to Healthy Elderly Subjects
The morphological correlation networks for both patients group (bvFTD and AD) were found to differ from the corresponding network for healthy elderly subjects in highly significant ways. Both groups showed a striking increase in the overall correlation strength in thickness and surface area networks compared with healthy elderly subjects. The effect was more pronounced in the cortical thickness network in all lobes for both positive and inverse correlations. This contrasts with a significantly lower correlation strength relative to normal for surface area in frontal lobe in AD and a directionally similar difference in bvFTD. This may be due to a larger number of correlations with a broad frequency distribution in disease as compared with sparser networks having a narrower frequency distribution in healthy elderly subjects. In addition to increased overall correlation strength, the number of within-lobe positive and inverse correlations as measured by node degree was higher in all lobes in both dementia groups than in healthy elderly controls. The number of between-lobe positive correlations in thickness as measured by the between-lobe participation index was also higher in all lobes. The number of within-lobe and between-lobe positive surface area correlations was also greater in both bvFTD and AD than in healthy elderly subjects in frontal, temporal, and parietal lobes. Both disease groups also differed from healthy elderly subjects in terms of the correlations in coupling between cortical thickness and surface area. Thus, both diseases are characterized by an overall increase in the strength and extent of structural correlation occurring both locally within lobes and globally between lobes.
The similarity between the two conditions in terms of the marked increase in overall strength and extent of structural correlation might appear to call into question the clinical distinctions between bvFTD and AD on which the classification of subjects in the study was based. Indeed, there were no differences between the two conditions in terms of overall cortical thickness and surface area. However, there were a number of important network differences between the two conditions. In the cortical thickness network, the overall positive correlation strength has been greater in bvFTD than in AD in frontal and temporal lobes, and the inverse correlation strength was also greater in bvFTD than in AD in the frontal lobe. The number of significant positive within-lobe correlations was higher in bvFTD than in AD in parietal and occipital lobes. Conversely, the number of positive and inverse within-lobe correlations was greater in AD than in bvFTD frontal and parietal lobes. Most of the inverse correlations in cortical thickness and surface area related to inter-hemispheric non-homologous fronto-temporal lobes in bvFTD and in fronto-parietal lobes in AD.
The hub-like organization of the correlation networks also differed substantially in the two conditions. Network connector hubs are though to provide network integration, whilst provincial hubs provide network segregation. It has been proposed that hubs provide resilience to insult in neurodegenerative disorders. Alternatively, it has been suggested that the hubs represent loci of particular vulnerability. It is therefore of interest to study how the hubs change in the context of neurodegenerative disease. bvFTD was characterized by an increase in the number of cortical thickness hubs in frontal lobe and a reduction or elimination of hubs in temporal, parietal, and occipital lobes. By contrast, AD was characterized by hubs distributed in all lobes, a reduction in the number of hubs in frontal cortex, and an increase in hubs in temporal and occipital lobes compared with bvFTD. In the positive correlation network for surface area, AD subjects had twice as many hubs overall than bvFTD, and the topology of these hubs differed. Thus overall, AD is characterized by a much more distributed pattern of hubs in both the thickness and surface area degenerative networks than bvFTD. By contrast, the hub-like organization is much more localized in bvFTD. It has been argued that bvFTD is a clinical syndrome with focal but heterogeneous atrophy centred around hubs. Identification of the insular region as one of the inverse network hubs (in both bvFTD and AD groups for the CT network) is consistent with the recent unexpected finding from diffusion MRI that there is an increase in hub-like fibre connectivity of the insula in bvFTD. The hubs of the healthy elderly group, on the other hand, were highly connected within and between lobes in a homologous fashion and were not otherwise linked to each other. The differences in hub-like organization between AD and bvFTD indicates differences in the hierarchy of nodal vulnerability and in the organization of network adaptations compensating differently in the two conditions. Thus, unlike lobar modularity, which is preserved in neurodegenerative disease, a constant hub-like organization is not preserved, implying that pre-existing hubs are not an intrinsic structural property of cortical network organization.
Although AD is also characterized by changes in cortical thickness, these are on the whole less marked than in bvFTD, whereas the changes in surface area are more prominent in AD, suggesting co-ordinated changes in numbers of adjacent affected columns. These differences would be consistent with the pathology of bvFTD affecting interneurons and astrocytes which have more localized links. The predominance of surface area correlations in AD would be consistent with the pathology affecting primarily long-tract cortico-cortical projection systems mediated by the principal cells. bvFTD differs in a number of important respects from AD: there is no cholinergic deficit in bvFTD, there is no treatment benefit from treatment with either acetylcholinesterase inhibiters or memantine, bvFTD is characterized by prominent astrocytic pathology, neurons affected in the neocortex are predominantly spiny interneurons in layers II and VI (pyramidal cells in layers III and V are predominantly affected in AD) and dentate gyrus of hippocampus (neurons affected in AD are in CA 1-4 and not dentate gyrus) and bvFTD is characterized by increased glutamate levels in the neocortex but AD is not. However, none of these conditions provide a simple explanation for the different distribution patterns of the correlated structural changes described herein.
Global Character and Significant of Network Changes in Dementia
The overall picture which emerges from the two disease groups studied is that network architecture is changed in a co-ordinated fashion throughout the whole brain as regards both positive and inverse correlations. This is surprising, given that the neurodegenerative processes in these two conditions are generally considered to be anatomically restricted, to frontal and temporal lobes in the case of bvFTD and to temporal and parietal lobes in AD. Rather, the network analysis suggests that there are changed in cortical thickness and surface area networks in both conditions that affect all lobes in a global manner, but that there are differences in the anatomical topology of the changes. Both Tau and TDP-43 aggregation pathology is known to spread in prion-like fashion, whereby pathology in an affected neuronal population can initial pathology in a connected, but previously unaffected neuronal population. The positive correlations could therefore reflect in part the spread of pathology in existing normal networks whereby existing functional networks are affected or spared together. Alternatively, such correlations might express functional dependencies, such that loss of function in one member of a partnership results in a parallel loss of function in a partner normally synchronized functionally with an affected node. This interpretation would be consistent with previous work on cortical thickness correlations in healthy adults, where positive correlations were found to converge with diffusion-based axonal connections.
The work discussed herein highlights for the first time the significance of inverse correlation networks. It should be noted that because the inverse correlations seen in both neurodegenerative disorders reflect primarily inter-lobar non-homologous associations, they would not have been detected using only a lobe-based approach to the analysis. It is particularly the appearance of these non-homologous inverse inter-lobar correlations and their increased strength that represents the clearest overall difference between neurodegenerative disease and normal aging. By contrast, the normal aging brain is characterized by substantially weaker homologous positive correlations. An attractive hypothesis is that, as certain nodes become functionally impaired, other still unaffected nodes compensate, accentuating non-homologous associations in disease. This would imply that the major reorganization in structural network observed may be partly adaptive in character. Structural plasticity has been demonstrated in other contexts, and functional compensation is known to occur in focal disease.
The work discussed herein represents a first comparative study of correlated structural network abnormalities in bvFTD and AD relative to healthy aging. These correlations arise from both positive and inversely linked changes in cortical thickness and surface area in the two disease conditions, which are quite different from those seen in normal elderly subjects. The changes seen in disease are global in character and are not restricted to fronto-temporal and temporo-parietal lobes respectively in bvFTD and AD. Rather, they appear to represent structural adaptations to neurodegeneration which differ in the two conditions. Further, all of the correlation networks showed a quite distinctive hub-like organization which differs both from normal and between the two forms of dementia. Unlike lobar organization of networks, which remains constant in disease, hub-like organization varies with the underlying pathology. This implies that hub-like organization is not a fixed feature of the brain and attempts to explain disease in terms of hubs may be inadequate. The differences between AD and bvFTD documented confirm that the clinical differences in the two dementia populations correspond to systematic differences in the underlying network structure of the cortex. The topological differences in thickness and surface-area hub-like organization, as well as the underlying positive and inverse correlation networks, may provide a basis for development of analytical tools to aid in the differential diagnosis in the two conditions, which can be difficult to distinguish by purely clinical criteria.
Use of Correlation Matrices in Determining Patient Group Response to Neuropharmacological Intervention
The methods discussed above have been used to determine patient group response to neuropharmacological intervention.
As can be seen from
These connections represent inverse correlations whereby a decrease in the volume or surface area of an affected area in a particular node (typically located in the posterior parts of the brain) is correlated in a statistically significant manner with a linked node where there is a corresponding increase in the volume or surface area. As discussed above, the presence of these non-homologous inverse correlations is indicative of neurodegenerative disease and most likely represent frontal compensation for posterior dysfunction arising from pathology. Symptomatic AD treatments induce an increase in these non-homologous compensatory linkages.
In summary, the structural correlation network analysis discussed above reveals the emergence of highly abnormal inverse non-homologous inter-lobar correlations in AD and bvFTD. It is hypothesized that these represent compensatory input from frontal brain regions unaffected or less affected by disease. Symptomatic treatments and LMTM act in fundamentally different ways in AD in terms of the structural correlation network. Symptomatic treatments induce substantial increases in the compensatory networks. LMTM as monotherapy reduces the need for these compensatory networks by reducing the primary pathology thereby permitting affected neurons to function more normally. These results confirm that the abnormal inverse non-homologous correlations seen in neurodegenerative diseases such as AD are adaptive in character, since they can be reversed or attenuated by disease-modifying treatment, but not by symptomatic AD treatments. The effects are seen in within-cohort before/after analysis in which subjects at baseline serve as their own controls for the changes occurring after receiving LMTM treatment at 8 mg/day as monotherapy for 65 weeks. These analyses are far more sensitive to treatment effects than crude whole brain or lobar volume analyses. Further, as will be discussed below, the results seen in terms of structural correlation networks are consistent with the functional effects seen by re-normalized partial directed coherence electroencephalography analysis techniques.
Structure/Function Correlation Using Electroencephalography (EEG)
Re-normalized partial directed coherence (rPDC) network approaches to EEG data allow an indication of the direction and strength of electrical activity within the brain to be investigated using a network approach. This is discussed, for example, in WO 2017/118733 (the entire contents of which is incorporated herein by reference).
The resulting network, as shown in
By counting the number of directed connections into and out of a given node and/or measuring their relative strength, it is possible to define whether a node is a sink (and has more and/or stronger connections in than out) or a source (and has more and/or stronger connections out than in). This is shown schematically in
After deriving the difference between incoming and outgoing connection for all nodes, it is possible then to provide a heat map indicative of the location and intensity of sinks and sources within a patient's brain. This may include a step of defining each node as either a sink or a source. n example of such a heat map is shown in
The methods discussed above were used to analyze data provided from 329 subjects divided into 167 diagnosed subjects (DS) and 162 paired volunteers (PV) at their initial assessment (visit 1):
As can be seen, diagnosed subjects are significantly more impaired cognitively on the MMSE and ADAS-Cog psychometric scales, and also have a higher score on the overall Clinical Dementia Rating (CDR) scale. Otherwise, there are no differences in age or sex distribution.
A machine learning classifier was trained on a set of the data provided by the 329 subjects discussed above. The β-band EEG data from 100 seconds of brain activity during eyes closed resting state was used in each case to prepare the rPDC network. The machine learning classifier was then used to classify all 329 subjects as either AD or paired volunteers (PV) achieving 95% accuracy. Moreover, the machine learning classifier can be used to estimate the probability that a subject has AD allowing for more than just a binary decision. For example, the subject whose heat map is shown in
Psychometric testing of the apparently healthy cohort showed downward cognitive trajectory on the Hopkins Verbal Learning Test over 18 months in a subset of the subjects. The characteristics of the cohort were as follows. As can be seen, there was no difference on cognitive score at baseline on the MMSE scale between those found to be at risk and those found not at risk of decline.
The heat map of the group of at risk subjects is shown in
As has been shown by the above, there are clear differences in networks between diagnosed subjects and paired volunteers. These differences are highly significant at group level. As will be appreciated, the first version of the machine learning classifier has a higher level of accuracy than routine superficial clinical assessment and gives probabilities of having AD at the individual subject level which can be used for decision making in further clinical management.
The increase in non-homologous inter-lobar compensatory inverse correlation networks seen in
There were also no statistically significant differences in ADAS-Cog or CDR scales.
As can be seen in
The frontal lobes show the same phenomenon by EEG as that shown by structural analyses of correlation networks in
The systems and methods of the above embodiments may be implemented in a computer system (in particular in computer hardware or in computer software) in addition to the structural components and user interactions described.
The term “computer system” includes the hardware, software and data storage devices for embodying a system or carrying out a method according to the above described embodiments. For example, a computer system may comprise a central processing unit (CPU), input means, output means and data storage. Preferably the computer system has a monitor to provide a visual output display. The data storage may comprise RAM, disk drives or other computer readable media. The computer system may include a plurality of computing devices connected by a network and able to communicate with each other over that network.
The methods of the above embodiments may be provided as computer programs or as computer program products or computer readable media carrying a computer program which is arranged, when run on a computer, to perform the method(s) described above.
The term “computer readable media” includes, without limitation, any non-transitory medium or media which can be read and accessed directly by a computer or computer system. The media can include, but are not limited to, magnetic storage media such as floppy discs, hard disc storage media and magnetic tape; optical storage media such as optical discs or CD-OMs; electrical storage media such as memory, including RAM, ROM and flash memory; and hybrids and combinations of the above such as magnetic/optical storage media.
While the invention has been described in conjunction with the exemplary embodiments described above, many equivalent modifications and variations will be apparent to those skilled in the art when given this disclosure. Accordingly, the exemplary embodiments of the invention set forth above are considered to be illustrative and not limiting. Various changes to the described embodiments may be made without departing from the spirit and scope of the invention.
In particular, although the methods of the above embodiments have been described as being implemented on the systems of the embodiments described, the methods and systems of the present invention need not be implemented in conjunction with each other, but can be implemented on alternative systems or using alternative methods respectively.
Gauthier, S. et al. “Efficacy and safety of tau-aggregation inhibitor therapy in patients with mild or moderate Alzheimer's disease: a randomized, controlled, double-blind, parallel-arm, phase 3 trial”, The Lancet 388, 2873-2884 (2016)
Wilcock, G. K. et al “Potential of low dose leuco-methylthioninium bis (hydromethanesulphonate) (lmtm) monotherapy for treatment of mild Alzheimer's disease: Cohort analysis as modified primary outcome in a phase iii clinical trial. Journal of Alzheimer's disease 61, 635-657 (2018)
Feldman, H. et al “A phase 3 trial of the tau and tdp-43 aggregation inhibitor, leuco-methylthioninium bis (hydromethanesulfonate) (lmtm), for behavioural variant frontotemporal dementia (bvFTD)” Journal of Neurochemistry 138, 255 (2016)
Murray, A. D. et al “The balance between cognitive reserve and brain imaging biomarkers of cerebrovascular and Alzheimer's diseases” Brain 134, 3687-3696 (2011)
Storey, J. D. “A direct approach to false discovery rates” Journal of the Royal Statistical Society: Series B (Statistical Methodology) 64, 479-498 (2002)
Van Wijk, B. C., Stam, C. J. & Daffertshofer, A. “Comparing brain networks of different size and connectivity density using graph theory.” PLoS One 5, e13701 (2010)
Rubinov M, Sporns O. “Weight-conserving characterization of complex functional brain networks”, Neuroimage, 56(4):2068-79 (2011)
All references referred to above are hereby incorporated by reference.
| Filing Document | Filing Date | Country | Kind |
|---|---|---|---|
| PCT/EP2018/073906 | 9/5/2018 | WO | 00 |
