METHOD TO DETERMINE COGNITIVE IMPAIRMENT

Abstract

A method of determining if a subject has a mild cognitive impairment (MCI) or Alzheimer's Disease, the method comprising measuring the concentration of one or more VOCs in an exhaled sample from a subject, and comparing the concentration with a reference concentration, I.

Claims

1. A method of determining if a subject has a mild cognitive impairment (MCI), the method comprising: a) measuring, in an exhaled sample from a subject, the concentration of one or both VOCs selected from the group consisting of hexanal and heptanal; b) comparing the concentration measured in a) with a reference concentration; and c) determining that if the concentration measured in a) is more than about 5% higher than the reference concentration, it is indicative that the subject has or will develop a MCI.

2. The method of claim 1, wherein (a) further comprises measuring, in an exhaled sample of the subject, the concentration of 2-propanol.

3. The method of claim 1 or claim 2, wherein determining that the concentration of hexanal measured in a) is more than about 100% higher than the reference concentration of hexanal is indicative that the subject has or will develop a MCI.

4. The method of claim 2, wherein determining that the concentration of 2-propanol in a) is more than about 10% higher than the reference concentration of 2-propanol is indicative that the subject has or will develop a MCI.

5. The method according to any one of the preceding claims, wherein determining that the concentration of heptanal in a) is more than about 10% higher than the reference concentration of heptanal is indicative that the subject has or will develop a MCI.

6. A method of determining if a subject has AD, the method comprising: b) measuring, in an exhaled sample from a subject, the concentration of one or both VOCs selected from the group consisting of acetone and 2-butanone; b) comparing the concentration measured in a) with a reference concentration; and c) determining that if the concentration measured in a) is more than about 5% higher or more than about 5% lower than in the reference concentration, it is indicative that the subject has or will develop AD.

7. The method according to claim 6, wherein (a) further comprises measuring, in an exhaled sample from a subject, the concentration of 2-propanol, optionally wherein determining that the concentration of 2-propanol measured in a) is higher than the reference concentration of 2-propanol is indicative that the subject has or will develop AD.

8. The method according to claim 6 or claim 7, wherein determining that the concentration of acetone measured in a) is higher than the reference concentration of acetone is indicative that the subject has or will develop AD, optionally wherein determining that the concentration of acetone measured in a) is more than about 5% higher than the reference concentration of acetone is indicative that the subject has or will develop AD.

9. The method according to claim 7, wherein determining that the concentration of 2-propanol measured in a) is more than about 10% higher than the reference concentration of 2-propanol is indicative that the subject has or will develop AD.

10. The method according to any one of claims 6 to 9, wherein determining that the concentration of 2-butanone measured in a) is lower than the reference concentration of 2-butanone is indicative that the subject has or will develop AD.

11. The method according to claim 10, wherein determining that the concentration of 2-butanone measured in a) is more than about 5% lower than the reference concentration of 2-butanone is indicative that the subject has or will develop AD.

12. A method of determining if a subject is has MCI or AD, the method comprising a) measuring, in an exhaled sample of a subject, the concentration of the VOC, 1-butanol; b) comparing the concentration measured in a) with a reference concentration; and c) determining that if the concentration measured in a) is about 5% to about 60% higher than the reference concentration, it is indicative that the subject has or will develop MCI, or determining that if the concentration measured in a) is more than about 60% higher than the reference concentration, it is indicative that the subject has or will develop AD.

13. The method according to claim 12, wherein (a) comprises measuring, in an exhaled sample of a subject, the concentration of 1-butanol and one or both VOCs selected from the group consisting of hexanal and 2-propanol.

14. The method according to claim 13, wherein determining that the concentration of hexanal measured in a) is about 5% to about 100% higher than the reference concentration of hexanal is indicative that the subject has or will develop AD.

15. The method according to claim 14, wherein determining that the concentration of 2-propanol measured in a) is about 5% to about 49% higher than the reference concentration of 2-propanol is indicative that the subject has or will develop MCI.

Description

FIGURE LEGENDS

[0073] For a better understanding of the invention, and to show how embodiments of the same may be carried into effect, reference will now be made, by way of example, to the accompanying Figures, in which:

[0074] FIG. 1 is a 3D topographic map generated by GC-IMS analysis of the constituents of an exhaled breath sample from a healthy individual; and

[0075] FIG. 2 is an AUC ROC curve based on the analysis of samples taken from different groups of subjects: Healthy vs MCI (blue); Healthy vs AD (black); MCI vs AD (red). Samples were analysed using a G.A.S. GC-IMS instrument.

EXAMPLES

[0076] The inventors demonstrated that exhaled volatile organic compounds (VOCs) in breath can be used as a non-invasive biomarker to distinguish healthy controls from MCI, healthy controls from AD, and MCI from AD. Detection of VOCs was performed using gas chromatography-ion mobility spectrometry (GC-IMS) techniques. Confounding factors, such as age, smoking habits, gender and alcohol consumption were also investigated to demonstrate the efficacy of results.

[0077] Materials and Methods

[0078] Subjects

[0079] A total of 100 subjects were recruited for this case-control study. Ethical approval was obtained from local research ethics committee (Ref No. 17/18-829, University of Plymouth, UK). MCI and AD patients were recruited by Re:Cognition Health (Plymouth, UK), along with their respective partners as healthy controls. The volunteers and patients received information sheets and were consented following a face-to-face interview by a medical doctor. The study cohort includes 50 patients (25 MCI, 25 AD) and 50 healthy control volunteers. MCI patients were recruited based on the investigator's assessment (inventor S.P.), who reviewed their clinical history and confirmed that the subject demonstrated amnestic symptoms highly suggestive of AD. Thus, MCI subjects were identified by clinical tests of mental function, in line with current practices defined by the NIAAA (National Institute of Aging-Alzheimer's Association, which defines national standards). Similar clinical standards were applied to the AD group as well. Specifically, AD subjects were recruited based on M-ACE scores below 23, a clinical assessment of the presence of amnestic problems, and another domain of cognitive problems, including patients with dementia who had functional impairment. Healthy controls included subjects that had no-known history of neurological disorders (self-reported). The mean age of the MCI and AD group was 74.9 (standard deviation: 7.6), including 29 males and 21 females. An overview of the demographic data of MCI, AD and control subjects is shown in Table 1.

TABLE-US-00001 TABLE 1 Demographic data of healthy controls and MCI and AD patients Parameter MCI (n = 25) AD (n = 25) Controls (n = 50) Mean age (SD) 72.2 (7.3) 77.5 (7.2) 71.2 (7.3) Gender ratio M:F 15:10 14:11 17:33 Smoking habits 16 ex-smokers 10 never smokers 22 never smokers 9 never smokers 12 ex-smokers 25 ex-smokers 3 current smokers 3 current smokers Alcohol - mean 10.8 (11.6) 3.5 (6.4) 11.3 (13.4) units/week (SD) Medication Omeprazole (8) Donepezil (10) N/A (No. of Subjects) Atorvastatin (6) Atorvastatin (6) Bisoprolol (5) Omeprazole (4) Paracetamol (4) Aspirin (4) Simvastatin (4) Amlodipine (4) Amlodipine (3) Folic Acid (4) Warfarin (3) Clopidogrel (3) Aspirin (3) Citalopram (3) Citalopram (3) 47 others (≤2) 64 others (≤2) Co-diseases/ Hypertension (11) Hypertension (7) N/A conditions Depression (6) High Cholesterol (5) (No. of Subjects) High Cholesterol (6) Gastric Reflux (4) Atrial Fibrillation (4) 54 others (≤2) Hiatus Hernia (3) Migraines (3) Diabetes Mellitus (3) 58 others (≤2)

[0080] Breath Analysis Platform

[0081] GC-IMS technology was used to detect VOCs in the exhaled breath of subjects. In recent years, there has been a growing presence of portable GC-IMS analysers, which have demonstrated capabilities in medical diagnostics. The BreathSpec (G.A.S., Dortmund, Germany) used, is a commercial instrument, consisting of a gas chromatograph (GC) and an ion mobility spectrometer (IMS). Importantly, the unit analyses at point of care, only takes a few minutes to undertake an analysis and is patient friendly, with breath samples collected without putting strain on the subject. The BreathSpec is equipped with a MXT-200 mid-polarity column (Thames Restek, Saunderton, UK) for gas chromatographic separation, based on chemical interactions with the column.

[0082] Thus the chemicals in an exhaled breath sample, including VOCs, are preseparated by the GC column based on their interaction with a retentive layer lining the inside of the column. Thus chemicals, including the VOCs, are eluted from the column at different times (known as the retention time). Following this, the separated chemicals (analytes) are further separated by IMS. The analytes are ionised using a radioactive source and injected into a drift tube, using a shutter grid. The ions drift against a buffer gas under influence of a uniform electric field (400 V/cm), where the various ions achieve different velocities, inversely related to their size, mass and charge. The ions are then collected on a Faraday plate, to produce a time-dependent signal that corresponds with ion mobility. The amount of time taken by an ion to reach the Faraday plate is referred to as the drift time. This technique can measure substances in the low parts-per-billion (ppb) range and delivers measurement results in less than 10 minutes. This unit recirculates and filters ambient air, which allows the unit to operate without the need of an external gas supply.

[0083] Breath Sampling

[0084] The sampling procedure requires only four seconds of exhaled breath. Subjects were provided with a disposable plastic mouthpiece, which pushes into the mouthpiece holder/sample inlet and connects directly to the front-panel of the instrument. The mouthpieces are open-ended, which allows air inside the mouthpiece to be displaced as exhalation proceeds. As a result, the sampling system can separate out the last portion of exhaled breath—known as end-tidal or alveolar breath. Alveolar breath refers to the last portion (350 mL) of exhaled breath, expelled from within the lungs and the lower-airways, which have undergone gaseous exchange with the blood in the alveoli. Users do not need to exhale until their lungs are as empty as possible and instead are simply asked to breathe normally. This improves reproducibility and makes the device suitable for vulnerable subjects, such as the elderly.

[0085] A typical GC-IMS output response to a breath sample (healthy control subject from this study), is shown in FIG. 1. The obtained sample is represented in a 3D topographic map, whereby each point is characterised by the retention time in the chromatographic column (in seconds), the drift time (in milliseconds) and the intensity of ion current signal (in millivolts). The signal intensity is indicated by colour, where each high-intensity area represents a single or combination of chemicals (with the same properties). The long line red line is the RIP (reactive ion peak), which is a background signal. Laboratory Analytical Viewer (LAV) software (v2.2.1, G.A.S., Dortmund, Germany) was used for GC-IMS signal viewing.

[0086] Data Analysis

[0087] The data analysis approach used focuses on distinguishing between the three diagnostic groups: AD, MCI and controls. However, since there are more controls than MCI or AD subjects, the effect of imbalanced datasets needs to be considered. For this reason, a random selection of 25 healthy subjects were used for this analysis. The first step of data analysis involves a pre-processing stage. The aim of this is to reduce the dimensionality, thus leaving data that is non-background. A typical GC-IMS dataset contains 11 million data points, which has high dimensionality, but low information content. Thus, the area of interest is cropped from the centre of the dataset and then a threshold is applied to remove the background. These steps reduce the number of data points by a factor of 100. Following this, a supervised feature selection procedure is undertaken, with class prediction performed using a k-fold cross-validation method (where k=10 in our case). This method involves partitioning the original data set into 10 equally-sized subsets. Of the 10 subsets, a single subset is retained as the validation data for testing the model, and remaining nine subsets are used as training data. Training features are identified using Wilcoxon rank-sum test between the two groups and those feature points with the lowest p-value selected and used to construct models based on five different classifiers. These features are identified purely on a statistical basis and not on any biological function at this stage. This analysis was run using R (version 3.6.0) with standard machine learning sub-packages: support vector machine (SVM)—kernlab; sparse logistic regression (SLR)—glmnet; Gaussian process—gbm, neural network—neuralnet, and random forest (RF)—randomForest. This process is repeated 10 times (number of folds), with each subset used once as validation data. The 10 results are then combined to produce a single estimation and from this, statistical parameters calculated.

[0088] In addition to the classification analysis, it is possible to identify unknown VOCs that relate significantly to the efficacy of the generated results. Using GC-IMS Library Search software (v1.0.1, G.A.S., Dortmund, Germany), we can potentially identify compounds based on gas chromatographic retention times and ion mobility drift times, by referencing a NIST database with around 83,000 compound entries. Here, the identified features are plotted back onto the original GC-IMS output and then the chemicals are identified. For quality control, the instrument was normalised to match the GC-IMS Library Search software with the equipped column using a standard ketone mix (2-butanone, 2-pentanone, 2-hexanone, 2-heptanone, 2-octanone and 2-nonanone).

[0089] Confounding Factors

[0090] Applications of exhaled breath analysis for diagnostic and/or monitoring purposes should consider possible confounding factors. These are factors which are known to have some impact on breath content and can thus introduce bias or generate spurious associations. For example, age is a critical confounder in this study, because increased age is the biggest risk factor for AD. In addition to age, smoking habits, gender and alcohol consumption were considered. The impact of these factors can be evaluated by re-running the classification analysis applied to the diagnostic groups, after re-organising the patients and volunteers based on confounding factors. To simplify the analysis and create more evenly-balanced groups, the confounding groups were defined as: Age [>75 vs <=74 years], smoking habits [never smokers vs ex/current smokers], gender [male vs female] and alcohol consumption [>=14 vs <14 units/week]. The latter threshold represents UK guidelines for regular alcohol consumption. The confounding factor groups are summarised in Table 2.

TABLE-US-00002 TABLE 2 Summary of confounding factor groups. Factor Groups MCI AD Healthy Total Age <=74 years 15 8 34 57 >75 years 10 17 16 43 Smoking Ex/current smokers 16 15 28 59 Never smokers 9 10 22 41 Gender Male 15 14 17 46 Female 10 11 33 54 Alcohol <14 units per week 22 14 31 67 >=14 units per week 3 11 19 33

[0091] Results

Example 1

Chemical Identification

[0092] VOC analysis indicates that the concentration of three compounds, identified as acetone, 2-propanol and 2-butanone play a crucial role in distinguishing between healthy controls and AD subjects. In the test for AD vs MCI, changes in the concentration of 2-propanol, hexanal and 1-butanol were significant. The separation of healthy controls and MCI relied on changes in the concentration of 2-propanol, hexanal and heptanal.

[0093] The observed changes, as measured by GC-IMS, are shown in table 3 with respect to all 100 subjects.

TABLE-US-00003 Controls (intensity of ion current signal in VOC AD MCI volts) Acetone ↑ 10.5% ↑↑ 48.4% 2.66 V 2-Butanone ↓ −11.4% ↓ −13.2% 2.33 V 2-Propanol ↑↑ 57.6% ↓ 39.9% 1.03 V Hexanal ↑ 52.4% ↑↑ 209.5% 0.27 V 1-Butanol ↑↑ 97.7% ↑ 11.8% 0.33 V Heptanal ↑ 20.9% ↑ 36.2% 0.06 V

Example 2

Statistical Analysis

[0094] Analysis results are presented in FIG. 2 as overlapping receiver operating characteristic (ROC) curves. The corresponding area under curve (AUC) is a measure of how well parameters can distinguish between diagnostic groups, i.e. MCI, AD and healthy. NPV (negative predictive value) and PPV (positive predictive value) for the different comparisons have also been calculated. Analysis results for diagnostic groups were achieved using SLR, and are shown in Table 4.

TABLE-US-00004 TABLE 4 GC-IMS diagnostic group results Test AUC ± 95% Sensitivity Specificity PPV NPV P-value Healthy 0.77 0.68 0.80 0.77 0.71 4.10 × vs MCI (0.64-0.90) (0.46-0.85) (0.59-0.93) 10.sup.−4 Healthy 0.83 0.60 0.96 0.94 0.71 3.48 × vs AD (0.72-0.94) (0.39-0.79) (0.80- 1.00) 10.sup.−5 MCI vs 0.70 0.60 0.84 0.79 0.68 0.0076 AD (0.55 -0.85) (0.39-0.79) (0.64-0.95)

Example 3

Confounding Factors

[0095] The analysis previously conducted on the AD, MCI and healthy groups was repeated, using the same analytical techniques and algorithms, on the confounding factor groups of age, smoking habits, gender and alcohol consumption. The analysis results are summarized in Table 5.

TABLE-US-00005 TABLE 5 GC-IMS confounding factors results Factor AUC ± 95% Sensitivity Specificity PPV NPV P-value Age 0.55 0.70 0.47 0.50 0.68 0.8070 [>75 vs (0.43-0.67) (0.54-0.83) (0.34-0.61) <=74 years] Alcohol 0.60 0.58 0.70 0.80 0.45 0.0492 [>=14 vs (0.48-0.72) (0.46-0.70) (0.51-0.84) <14 units/week] Gender 0.54 0.70 0.46 0.52 0.64 0.2668 [Male vs (0.42-0.65) (0.54-0.82) (0.33-0.60) Female] Smoking 0.56 0.81 0.39 0.66 0.59 0.1515 [Ex/current (0.44-0.68) (0.69-0.90) (0.24-0.55) vs Never]

[0096] Table 5 demonstrates that the possible confounding factors of gender, smoking and age have insignificant influence on breath content. However, alcohol consumption seems to have the most influence on breath, with an AUC of 0.60.

[0097] Discussion

[0098] While the exact mechanisms related to AD pathogenesis are not fully understood, there is some evidence to suggest that defects in mitochondrial metabolism (i.e. changes in mitochondrial function and potential dysfunction) play a key role in neurodegeneration. Furthermore, it is believed that mitochondrial dysfunctions in NDDs are associated with increased production of reactive oxygen species (ROS), which cause cell damage and intercellular oxidative stress. Endogenous VOCs (produced in the body) follow metabolic pathways and are transported via the bloodstream to the lungs, where they are exhaled in breath. Oxidative stress has been detected in blood and thus presents an opportunity for the application of breath analysis to facilitate the discovery and evaluation of biomarkers associated with cellular energy metabolism, mitochondrial dysfunction and oxidative stress.

[0099] A key advantage of the results within this application is that the ages and genders of the subject cohort are approximately balanced. Furthermore, by recruiting MCI and AD patients with their respective partners as healthy controls, it is possible to design a more robust experiment and minimise the possible effects of lifestyle and environmental factors, which can potentially introduce significant interpersonal differences. Moreover, age-matching is a critical factor to consider, since more women than men have AD and other forms of dementia. For example, almost two-thirds of Americans with Alzheimer's are women. It has been suggested that this discrepancy is due to women generally living longer than men, which increases the risk factor of developing AD. However, there are other factors, such as sex-specific genetic and hormonal factors which can contribute to variance in clinical efficacy. Moreover, lifestyle choices such as smoking, excessive alcohol consumption, poor diet and resulting health conditions (obesity, type-2 diabetes, and cardiovascular disease) can have varied impacts on dementia risk, depending on sex. For this analysis, age was subdivided into ‘young elderly’ (aged 65-74) and ‘older elderly’ (aged 75 years or older). This division accounts for the sharp rise in dementia cases over the age of 75-15% of those with Alzheimer's are aged 65-75, while 44% are aged 75-85.

[0100] VOC analysis indicates that acetone, 2-propanol and 2-butanone contributed significantly to the efficacy of our analysis for AD vs controls. These compounds are generally associated with normal breath. Similarly, changes in other breath markers, such as 2-propanol, hexanal, heptanal and 1-butanol, contributed to separation in AD vs MCI and MCI vs control tests. This suggests that AD-related changes in breath have a subtle impact on overall breath content. Changes in acetone are of particular interest. Early stages of AD are associated with region-specific declines in brain glucose metabolism. This can be supplemented by ketone bodies, including acetoacetate, β-hydroxybutyrate and acetone. These are usually produced from fat stores when glucose is unavailable, e.g. during prolonged fasting or when subscribing to a ketogenic diet, which could lead to increased levels of acetone in the exhaled breath in AD patients. AD patients often suffer loss of appetite, with nearly half of mild-AD subjects reporting appetite changes. The changes observed in exhaled acetone in this study could be related to this phenomenon. To the best of our knowledge, there is currently no known link between 1-butanol and metabolic pathways relating to AD or other NDDs.

[0101] Analysis of possible confounding factors indicate that gender, smoking, age and alcohol consumption have insignificant influence of breath content. Of these factors, alcohol seems to have the most influence, with an AUC of around 0.60. However, this factor is not significant enough to create two distinct groups or undermine the AD-related analysis.

CONCLUSIONS

[0102] Results from this study confirm the potential utility of analysing breath VOCs to distinguish between MCI, AD and healthy controls. Though this was a simple study, with relatively clean/well defined groups, the approach used was consistently able to separate between diagnostic groups [AUC±95%, sensitivity, specificity], Healthy vs MCI: [0.77 (0.64-0.9), 0.68, 0.8], Healthy vs AD: [0.83 (0.72-0.94), 0.6, 0.96], and MCI vs AD: [0.70 (0.55-0.85), 0.6, 0.84]. Analysis of possible confounding factors suggest that gender, age, smoking habits and alcohol consumption had insignificant influence on breath content. VOC analysis indicates that six compounds, tentatively identified as acetone, 2-propanol, 2-butanone, hexanal, heptanal and 1-butanol play a crucial role in distinguishing between diagnostic groups. The GC-IMS analysis technique was shown to be suitable for non-invasive sampling of elderly subjects and demonstrates potential as a fast, high-throughput, real-time diagnostic tool for AD in a point-of-care clinical setting.