Australian Imaging Biomarkers Research Articles

Elucidating the Link Between Anxiety/Depression and Alzheimer’s Dementia in the Australian Imaging Biomarkers and Lifestyle (AIBL) Study

BackgroundThe associations between mood disorders (anxiety and depression) and mild cognitive impairment (MCI) or Alzheimer’s dementia (AD) remain unclear.MethodsData from the Australian Imaging, Biomarker & Lifestyle (AIBL) study were subjected to logistic regression to determine both cross-sectional and longitudinal associations between anxiety/depression and MCI/AD. Effect modification by selected covariates was analysed using the likelihood ratio test.ResultsCross-sectional analysis was performed to explore the association between anxiety/depression and MCI/AD among 2,209 participants with a mean [SD] age of 72.3 [7.4] years, of whom 55.4% were female. After adjusting for confounding variables, we found a significant increase in the odds of AD among participants with two mood disorders (anxiety: OR 1.65 [95% CI 1.04–2.60]; depression: OR 1.73 [1.12–2.69]). Longitudinal analysis was conducted to explore the target associations among 1,379 participants with a mean age of 71.2 [6.6] years, of whom 56.3% were female. During a mean follow-up of 5.0 [4.2] years, 163 participants who developed MCI/AD (refer to as PRO) were identified. Only anxiety was associated with higher odds of PRO after adjusting for covariates (OR 1.56 [1.03–2.39]). However, after additional adjustment for depression, the association became insignificant. Additionally, age, sex, and marital status were identified as effect modifiers for the target associations.ConclusionOur study provides supportive evidence that anxiety and depression impact on the evolution of MCI/AD, which provides valuable epidemiological insights that can inform clinical practice, guiding clinicians in offering targeted dementia prevention and surveillance programs to the at-risk populations.

Disease-driven domain generalization for neuroimaging-based assessment of Alzheimer's disease.

Development of deep learning models to evaluate structural brain changes caused by cognitive impairment in MRI scans holds significant translational value. The efficacy of these models often encounters challenges due to variabilities arising from different data generation protocols, imaging equipment, radiological artifacts, and shifts in demographic distributions. Domain generalization (DG) techniques show promise in addressing these challenges by enabling the model to learn from one or more source domains and apply this knowledge to new, unseen target domains. Here we present a framework that utilizes model interpretability to enhance the generalizability of classification models across various cohorts. We used MRI scans and clinical diagnoses from four independent cohorts: Alzheimer's Disease Neuroimaging Initiative (ADNI, n = 1821), the Framingham Heart Study (FHS, n = 304), the Australian Imaging Biomarkers & Lifestyle Study of Ageing (AIBL, n = 661), and the National Alzheimer's Coordinating Center (NACC, n = 4647). With this data, we trained a deep neural network to focus on areas of the brain identified as relevant to the disease for model training. Our approach involved training a classifier to differentiate between structural neurodegeneration in individuals with normal cognition (NC), mild cognitive impairment (MCI), and dementia due to Alzheimer's disease (AD). This was achieved by aligning class-wise attention with a unified visual saliency prior, which was computed offline for each class using all the training data. Our method not only competes with state-of-the-art approaches but also shows improved correlation with postmortem histology. This alignment with the gold standard evidence is a significant step towards validating the effectiveness of DG frameworks, paving the way for their broader application in the field.

Distance-based novelty detection model for identifying individuals at risk of developing Alzheimer's disease.

Novelty detection (ND, also known as one-class classification) is a machine learning technique used to identify patterns that are typical of the majority class and can discriminate deviations as novelties. In the context of Alzheimer's disease (AD), ND could be employed to detect abnormal or atypical behavior that may indicate early signs of cognitive decline or the presence of the disease. To date, few research studies have used ND to discriminate the risk of developing AD and mild cognitive impairment (MCI) from healthy controls (HC). In this work, two distinct cohorts with highly heterogeneous data, derived from the Australian Imaging Biomarkers and Lifestyle (AIBL) Flagship Study of Ageing project and the Fujian Medical University Union Hospital (FMUUH) China, were employed. An innovative framework with built-in easily interpretable ND models constructed solely on HC data was introduced along with proposing a strategy of distance to boundary (DtB) to detect MCI and AD. Subsequently, a web-based graphical user interface (GUI) that incorporates the proposed framework was developed for non-technical stakeholders. Our experimental results indicate that the best overall performance of detecting AD individuals in AIBL and FMUUH datasets was obtained by using the Mixture of Gaussian-based ND algorithm applied to single modality, with an AUC of 0.8757 and 0.9443, a sensitivity of 96.79% and 89.09%, and a specificity of 89.63% and 90.92%, respectively. The GUI offers an interactive platform to aid stakeholders in making diagnoses of MCI and AD, enabling streamlined decision-making processes. More importantly, the proposed DtB strategy could visually and quantitatively identify individuals at risk of developing AD.

Exploring the association between cancer and cognitive impairment in the Australian Imaging Biomarkers and Lifestyle (AIBL) study

An inverse association between cancer and Alzheimer’s disease (AD) has been demonstrated; however, the association between cancer and mild cognitive impairment (MCI), and the association between cancer and cognitive decline are yet to be clarified. The AIBL dataset was used to address these knowledge gaps. The crude and adjusted odds ratios for MCI/AD and cognitive decline were compared between participants with/without cancer (referred to as C+ and C− participants). A 37% reduction in odds for AD was observed in C+ participants compared to C− participants after adjusting for all confounders. The overall risk for MCI and AD in C+ participants was reduced by 27% and 31%, respectively. The odds of cognitive decline from MCI to AD was reduced by 59% in C+ participants after adjusting for all confounders. The risk of cognitive decline from MCI to AD was halved in C+ participants. The estimated mean change in Clinical Dementia Rating-Sum of boxes (CDR-SOB) score per year was 0.23 units/year higher in C− participants than in C+ participants. Overall, an inverse association between cancer and MCI/AD was observed in AIBL, which is in line with previous reports. Importantly, an inverse association between cancer and cognitive decline has also been identified.

Unpuzzling the relationship between hypertension, stroke, and Alzheimer’s disease – an observation from the Australian Imaging Biomarkers and Lifestyle Study

AbstractBackgroundAlzheimer’s disease (AD), hypertension and stroke are common diseases in elderly people, but whether they are intrinsically linked with each other remains uncertain. The aim of this study was to investigate the risks of people who have hypertension and stroke to develop mild cognitive impairment (MCI) and AD, and if hypertension and stroke are risk factors for AD.MethodThe longitudinal data was retrieved from the Australian Imaging Biomarkers and Lifestyle Study (AIBL). The prevalence of hypertension and stroke was compared between cognitively normal (CN) participants, participants who were CN but subsequently diagnosed with MCI/AD during the study (refer to as participants exhibited cognitive decline), and participants with AD confirmed at recruitment.ResultThe prevalence of hypertension (odds ratio [OR] = 1.37, 95% confidence interval [CI] = 1.05 ‐ 1.77, p = 0.018) was significantly higher in participants exhibited cognitive decline than CN participants. Interestingly, the prevalence of hypertension (OR = 0.88, 95% CI = 0.70 ‐ 1.11, p = 0.287) was comparable between CN and AD participants. This observation suggested that hypertension is a risk factor for the progression of cognitive decline. The prevalence of stroke was significantly higher in participants exhibited cognitive decline (OR = 2.04, 95% CI = 1.10 ‐ 3.79, p = 0.024) and AD participants (OR = 3.43, 95% CI = 2.09 – 5.64, p<0.001) than CN participants. This observation suggested that stroke is a risk factor for the progression of cognitive decline and AD.ConclusionThis preliminary study demonstrated that hypertension and stroke are risk factors for the progression of cognitive decline, and stroke is a risk factor for AD. Given the elevated risk in cognitive decline, the cognitive function of elderly people with hypertension or history of stroke should be monitored for timely detection of MCI and AD.

Apolipoprotein E Receptor 2 is associated with Alzheimer’s disease pathogenesis

AbstractBackgroundGenetic variations to apolipoprotein E (apoE) underlie the strongest risk for Alzheimer´s disease (AD) after aging but it is not yet known how apoE confers risk for AD. While most research is focusing on dysregulations associated with the presence of the risk allele of apoE, several aspects of apoE biology remain poorly understood. In particular, the physiological relevance of apoE interaction with members of the low‐density lipoprotein (LDL) receptor gene family is not well understood.MethodFerroptosis was induced in the N27 neuronal cell model of ferroptosis using multiple agents (erastin, RSL3, iron, cysteine depletion). Genetic variants of apolipoprotein E receptor 2 (apoER2) were generated via site‐directed mutagenesis and expressed in N27 neurons. The severity of ferroptosis was measured by several toxicity assays (MTT, LDH) and lipid peroxide probe (C11BODIPY). The association between apoER2 polymorphism and cognition were investigated in two longitudinal studies: the Alzheimer’s Disease Neuroimaging Initiative (ADNI) and the Australian Imaging Biomarkers and Lifestyle Study of Ageing (AIBL).ResultWe recently reported that apoE is a potent inhibitor of ferroptosis in neuronal cells in a mechanism that involves a receptor‐mediated activation of the PI3K/AKT signaling pathway. In this study, we identified genetic variants of apoER2 (also known as LRP8, a member of the LDL receptor family) that were associated with slower cognitive decline in clinical cohorts of AD. In preliminary work, we show that one apoER2 variant was associated with an increased protection from ferroptosis in an apoE‐dependent process.ConclusionWe present evidence for a biological role of apoER2 and its genetic variation on pathologies associated with Alzheimer’s disease, which may help develop new diagnostic and therapeutic approaches towards the disease.

Developing a classifier to screen for mild cognitive impairment and Alzheimer’s disease using machine learning approaches

AbstractBackgroundMachine learning classifier is a powerful tool that has been recently applied to datasets of Alzheimer’s disease (AD). The majority of approaches developed to classify people with mild cognitive impairment (MCI) and Alzheimer’s disease (AD) from cognitively normal (CN) people have made use of expensive or invasive measurements, discouraging the clinical use of these tools. In the current study, we aimed to develop a classifier to screen for MCI and AD using data that can be routinely collected at low cost (e.g. blood pressure, ApoE genotype).MethodDatabase of the Australian Imaging Biomarkers and Lifestyle Study of Ageing (AIBL) was used (n = 14553 data entries). Dataset was split into two parts: training (80%) and testing (20%). A decision tree was first employed to screen for factors (i.e. clinical parameters) that can effectively discriminate CN, MCI, and AD (Step 1). To further refine factor selection, the association between the factors (from screening) and health status (CN, MCI, AD) were assessed using linear and logistic regression methods. Factors that were rejected by the hypothesis test (i.e. non‐obvious relationship with MCI or AD) and those require high cost for collection were removed (Step 2). The remaining factors were used for the construction of the classifier, where decision tree, Gaussian Naïve Bayes and second‐order polynomial approaches were employed.ResultsThe selected factors for classifier construction were ApoE genotype (p<0.001), medical history of stroke (p<0.001 AD, p = 0.223 MCI), diabetes (p<0.001) and depression (p<0.001), and systolic blood pressure (p<0.001). Three models were used to construct the classifier, and the differential accuracy ranges between 74.1 – 84.2% for CN, 25.3 – 50.2% for MCI, and 47.8 – 73.4% for AD (Table 1). Accuracy from second‐order polynomial model was lower than that from the decision tree and Gaussian Naïve Bayes model.ConclusionThe classifier developed in the current study is yet to be optimized, especially for MCI. In future study, the performance of our classifier will be evaluated using other AD datasets for generalizability. This classifier can potentially be used by the clinicians as an MCI/AD screening tool for at‐risk population prior to requesting costly diagnostic tests.

In-house data adaptation to public data: Multisite MRI harmonization to predict Alzheimer’s disease conversion

For the machine learning-based prediction of the conversion from mild cognitive impairment to Alzheimer’s disease, the collection of sufficient data to train a model is required, which involves a lot of time and expense. When data is not enough, combining public and in-house data may be appropriate by applying domain adaptation that alleviates inter-site heterogeneity. Existing methods simultaneously transform in-house and public data to represent them into a common feature space, and then train a classifier using labels in public data. However, this procedure causes the time- and cost-consuming re-training of classifier whenever in-house data changes, and also inheres the risk of information loss in public data. Motivated by this, we propose a method that only transforms in-house data while preserving public data, namely one-way domain adaptation. The proposed method represents in-house data similar with public data by matching the data distribution and the connectivity between brain regions with mean vectors and covariance matrices, respectively. Then, the pre-trained classifier in public data is applied to predict AD conversion for in-house data. The experiments, which use the Australian Imaging Biomarkers and Lifestyle Study of Aging and the Open Access Series of Imaging Studies as the in-house data and the Alzheimer's Disease Neuroimaging Initiative as the public data, show the effectiveness and efficiency of the proposed method, improving prediction performance about 34.8% on average without labels in the in-house datasets.

Cascaded Multi-Modal Mixing Transformers for Alzheimer’s Disease Classification with Incomplete Data

Accurate medical classification requires a large number of multi-modal data, and in many cases, different feature types. Previous studies have shown promising results when using multi-modal data, outperforming single-modality models when classifying diseases such as Alzheimer’s Disease (AD). However, those models are usually not flexible enough to handle missing modalities. Currently, the most common workaround is discarding samples with missing modalities which leads to considerable data under-utilisation. Adding to the fact that labelled medical images are already scarce, the performance of data-driven methods like deep learning can be severely hampered. Therefore, a multi-modal method that can handle missing data in various clinical settings is highly desirable. In this paper, we present Multi-Modal Mixing Transformer (3MT), a disease classification transformer that not only leverages multi-modal data but also handles missing data scenarios. In this work, we test 3MT for AD and Cognitively normal (CN) classification and mild cognitive impairment (MCI) conversion prediction to progressive MCI (pMCI) or stable MCI (sMCI) using clinical and neuroimaging data. The model uses a novel Cascaded Modality Transformers architecture with cross-attention to incorporate multi-modal information for more informed predictions. We propose a novel modality dropout mechanism to ensure an unprecedented level of modality independence and robustness to handle missing data scenarios. The result is a versatile network that enables the mixing of arbitrary numbers of modalities with different feature types and also ensures full data utilization in missing data scenarios. The model is trained and evaluated on the Alzheimer’s Disease Neuroimaging Initiative (ADNI) dataset with the state-of-the-art performance and further evaluated with The Australian Imaging Biomarker & Lifestyle Flagship Study of Ageing (AIBL) dataset with missing data.

A universal neocortical mask for Centiloid quantification.

The Centiloid (CL) project was developed to harmonize the quantification of amyloid beta (Aβ) positron emission tomography (PET) scans to a unified scale. The CL neocortical mask was defined using 11C Pittsburgh compound B (PiB), overlooking potential differences in regional distribution among Aβ tracers. We created a universal mask using an independent dataset of five Aβ tracers, and investigated its impact on inter-tracer agreement, tracer variability, and group separation. Using data from the Alzheimer's Dementia Onset and Progression in International Cohorts (ADOPIC) study (Australian Imaging Biomarkers and Lifestyle + Alzheimer's Disease Neuroimaging Initiative + Open Access Series of Imaging Studies), age-matched pairs of mild Alzheimer's disease (AD) and healthy controls (HC) were selected: 18F-florbetapir (N=147 pairs), 18F-florbetaben (N=22), 18F-flutemetamol (N=10), 18F-NAV (N=42), 11C-PiB (N=63). The images were spatially and standardized uptake value ratio normalized. For each tracer, the mean AD-HC difference image was thresholded to maximize the overlap with the standard neocortical mask. The universal mask was defined as the intersection of all five masks. It was evaluated on the Global Alzheimer's Association Interactive Network (GAAIN) head-to-head datasets in terms of inter-tracer agreement and variance in the young controls (YC) and on the ADOPIC dataset comparing separation between HC/AD and HC/mild cognitive impairment (MCI). In the GAAIN dataset, the universal mask led to a small reduction in the variance of the YC, and a small increase in the inter-tracer agreement. In the ADOPIC dataset, it led to a better separation between HC/AD and HC/MCI at baseline. The universal CL mask led to an increase in inter-tracer agreement and group separation. Those increases were, however, very small, and do not provide sufficient benefits to support departing from the existing standard CL mask, which is suitable for the quantification of all Aβ tracers. This study built an amyloid universal mask using a matched cohort for the five most commonly used amyloid positron emission tomography tracers.There was a high overlap between each tracer-specific mask.Differences in quantification and group separation between the standard and universal mask were small.The existing standard Centiloid mask is suitable for the quantification of all amyloid beta tracers.

Centiloid harmonisation across PET cameras without using paired scan

AbstractBackgroundThe Centiloid (CL) scale allows quantification of amyloid‐b accumulation across multiple PET tracers for the diagnosis of Alzheimer’s disease. The CL quantification on different PET cameras may result in significant quantification bias when merging cohorts acquired from multiple cameras or in the longitudinal studies when change of camera occurs. In this study, we propose a harmonisation method, based on the non‐negative matrix factorisation (NMF), to reduce the quantification bias between different cameras without using paired scans.MethodThree hundred and forty‐eight 18F‐NAV4694 scans from Australian Imaging Biomarkers and Lifestyle study acquired on three PET cameras (N = 116 each), Siemens Biograph mCT, Philips Allegro and Philips Gemini TF64 were used to train the model. Participants who switched camera between two consecutive timepoints were selected as the validation data. Seventeen participants imaged on the Biograph mCT and Gemini (2.63±0.51 years scan interval), and fifty‐two participants imaged on the Gemini and Allegro (2.73±0.73 years). One hundred participants who have two scans (2.26±0.82 years) on Gemini were used to define a reference rate of change (CL/year). Figure 1 illustrates the harmonisation workflow.ResultFigure 2 compares the curves of CL change rate against baseline CL for mCT to Gemini compared to the reference. The harmonised fitting is closer to the reference than the one without harmonisation in the low to middle CL range. The fitting in high CL range (CL>80) deviates from the reference probably because of the lack of subjects in this range. Figure 3 compares the results between Phillips Allegro and Gemini TF 64. The harmonised fitting in Figure 3 is nearly parallel to the reference for the entire CL range. In contrast, the fitting on the unharmonized data shows negative rate of change for the very negative subjects and much higher CL change rate for the very positive subjects, which are strongly different compared to the reference.ConclusionWe propose a practical and economical CL harmonisation for different PET cameras without using paired data. The NMF based method shows promising results for scanner harmonisation, though more validation are needed in future for different PET models.

Classification of amyloid positivity in PET imaging using end‐to‐end deep learning: a multi‐cohort, multi‐tracer analysis

AbstractBackgroundPositron emission tomography (PET) imaging with amyloid‐specific tracers is the gold standard for assessing amyloid positivity in vivo. However, visual interpretation of PET images is limited by availability of clinicians and interobserver variability. Classification of amyloid status typically requires calculation of standardized uptake value ratios (SUVRs), correspondence with neuroanatomy from magnetic resonance imaging, and additional complex, computationally intensive preprocessing steps requiring specialized software. We developed a deep learning model that can automatically classify unprocessed PET images as amyloid positive or negative.MethodFlorbetapir (AV45), florbetaben (FBB), and Pittsburgh compound B (PiB) images from the Alzheimer’s Disease Neuroimaging Association (ADNI), AV45 and PiB images from the Open Access Series of Imaging Studies 3 (OASIS3), and FBB, flutametamol (FMT), and flutafuranol (NAV4694) images from the Australian Imaging Biomarkers and Lifestyle Study of Ageing (AIBL) were used (Table 1). Each image was resized to isotropic 2 mm voxels, and voxel intensities between the 5th and 95th percentile were normalized to zero mean and unit variance. We trained and internally validated a convolutional neural network (CNN) on AV45 images from ADNI before testing it on images from an independent cohort and/or acquired with a different tracer.ResultThe model achieved an AUC of 0.98 [0.97, 1.0] on test AV45 images from ADNI and generalized well to AV45 images from OASIS3, achieving an AUC of 0.95 [0.92, 0.97] (Table 2). Critically, the model’s performance remained robust even when evaluating imaging from different tracers, including FBB, FMT, PiB, and NAV4694, scoring an AUC of at least 0.97 on each and demonstrating its ability to learn a tracer‐agnostic pattern of abnormality. Similar conclusions can be drawn from other performance metrics.ConclusionWe developed a deep learning model that can classify amyloid PET scans with a high degree of accuracy. We demonstrated that the model generalizes well to unseen images acquired from an independent cohort, under a different protocol, and/or with tracers unseen during training. Our model may usefully provide concise metrics for amyloid research as well as support clinical assessment of amyloid‐specific PET when expertise or computational resources are limited.

Tau, β-Amyloid, and Glucose Metabolism Following Service-Related Traumatic Brain Injury in Vietnam War Veterans: The Australian Imaging Biomarkers and Lifestyle Study of Aging-Veterans Study (AIBL-VETS).

Traumatic brain injury (TBI) is common among military veterans and has been associated with an increased risk of dementia. It is unclear if this is due to increased risk for Alzheimer's disease (AD) or other mechanisms. This case control study sought evidence for AD, as defined by the 2018 National Institute on Aging - Alzheimer's Association (NIA-AA) research framework, by measuring tau, β-amyloid, and glucose metabolism using positron emission tomography (PET) in veterans with service-related TBI. Seventy male Vietnam war veterans-40 with TBI (age 68.0 ± 2.5 years) and 30 controls (age 70.1 ± 5.3 years)-with no prior diagnosis of dementia or mild cognitive impairment underwent β-amyloid (18F-Florbetaben), tau (18F-Flortaucipir), and fluorodeoxyglucose (18F-FDG) PET. The TBI cohort included 15 participants with mild, 16 with moderate, and nine with severe injury. β-Amyloid level was calculated using the Centiloid (CL) method and tau was measured by standardized uptake value ratios (SUVRs) using the cerebellar cortex as reference region. Analyses were adjusted for age and APOE-e4. The findings were validated in an independent cohort from the Department of Defense-Alzheimer's Disease Neuroimaging Initiative (DOD ADNI) study. There were no significant nor trending differences in β-amyloid or tau levels or 18F-FDG uptake between the TBI and control groups before and after controlling for covariates. The β-amyloid and tau findings were replicated in the DOD ADNI validation cohort and persisted when the Australian Imaging Biomarkers and Lifestyle study of aging-Veterans study (AIBL-VETS) and DOD ADNI cohorts were combined (114 TBI vs. 87 controls in total). In conclusion, no increase in the later life accumulation of the neuropathological markers of AD in veterans with a remote history of TBI was identified.

Habitual nitrate intake and cognition in the Australian Imaging Biomarkers and Lifestyle study of ageing

An abstract is not available for this content. As you have access to this content, full HTML content is provided on this page. A PDF of this content is also available in through the ‘Save PDF’ action button.

Classification of amyloid positivity in PET imaging using end‐to‐end deep learning: a multi‐cohort, multi‐tracer analysis

AbstractBackgroundPositron emission tomography (PET) imaging with amyloid‐specific tracers is the gold standard for assessing amyloid positivity in vivo. However, visual interpretation of PET images is limited by availability of clinicians and interobserver variability. Classification of amyloid status typically requires calculation of standardized uptake value ratios (SUVRs), Correspondence with neuroanatomy from magnetic resonance imaging, and additional complex, computationally intensive preprocessing steps requiring specialized software. We developed a deep learning model that can automatically classify unprocessed PET images as amyloid positive or negative.MethodFlorbetapir (AV45), florbetaben (FBB), and Pittsburgh compound B (PiB) images from the Alzheimer’s Disease Neuroimaging Association (ADNI), AV45 and PiB images from the Open Access Series of Imaging Studies 3 (OASIS3), and FBB, flutametamol (FMT), and flutafuranol (NAV4694) images from the Australian Imaging Biomarkers and Lifestyle Study of Ageing (AIBL) were used (Table 1). Each image was resized to isotropic 2 mm voxels, and voxel intensities between the 5th and 95th percentile were normalized to zero mean and unit variance. We trained and internally validated a convolutional neural network (CNN) on AV45 images from ADNI before testing it on images from an independent cohort and/or acquired with a different tracer.ResultThe model achieved an AUC of 0.98 [0.97, 1.0] on test AV45 images from ADNI and generalized well to AV45 images from OASIS3, achieving an AUC of 0.95 [0.92, 0.97] (Table 2). Critically, the model’s performance remained robust even when evaluating imaging from different tracers, including FBB, FMT, PiB, and NAV4694, scoring an AUC of at least 0.97 on each and demonstrating its ability to learn a tracer‐agnostic pattern of abnormality. Similar conclusions can be drawn from other performance metrics.ConclusionWe developed a deep learning model that can classify amyloid PET scans with a high degree of accuracy. We demonstrated that the model generalizes well to unseen images acquired from an independent cohort, under a different protocol, and/or with tracers unseen during training. Our model may usefully provide concise metrics for amyloid research as well as support clinical assessment of amyloid‐specific PET when expertise or computational resources are limited.

Centiloid harmonisation across PET cameras without using paired scan

AbstractBackgroundThe Centiloid (CL) scale allows quantification of amyloid‐b accumulation across multiple PET tracers for the diagnosis of Alzheimer’s disease. The CL quantification on different PET cameras may result in significant quantification bias when merging cohorts acquired from multiple cameras or in the longitudinal studies when change of camera occurs. In this study, we propose a harmonisation method, based on the non‐negative matrix factorisation (NMF), to reduce the quantification bias between different cameras without using paired scans.MethodThree hundred and forty‐eight 18F‐NAV4694 scans from Australian Imaging Biomarkers and Lifestyle study acquired on three PET cameras (N=116 each), Siemens Biograph mCT, Philips Allegro and Philips Gemini TF64 were used to train the model. Participants who switched camera between two consecutive timepoints were selected as the validation data. Seventeen participants imaged on the Biograph mCT and Gemini (2.63±0.51 years scan interval), and fifty‐two participants imaged on the Gemini and Allegro (2.73±0.73 years). One hundred participants who have two scans (2.26±0.82 years) on Gemini were used to define a reference rate of change (CL/year). Figure 1 illustrates the harmonisation workflow.ResultFigure 2 compares the curves of CL change rate against baseline CL for mCT to Gemini compared to the reference. The harmonised fitting is closer to the reference than the one without harmonisation in the low to middle CL range. The fitting in high CL range (CL>80) deviates from the reference probably because of the lack of subjects in this range. Figure 3 compares the results between Phillips Allegro and Gemini TF 64. The harmonised fitting in Figure 3 is nearly parallel to the reference for the entire CL range. In contrast, the fitting on the unharmonized data shows negative rate of change for the very negative subjects and much higher CL change rate for the very positive subjects, which are strongly different compared to the reference.ConclusionWe propose a practical and economical CL harmonisation for different PET cameras without using paired data. The NMF based method shows promising results for scanner harmonisation, though more validation are needed in future for different PET models.

Objectively measured physical activity and cognition in cognitively normal older adults: A longitudinal analysis of the Australian Imaging Biomarkers and Lifestyle (AIBL) study

AbstractBackgroundPhysical inactivity is one of the greatest modifiable risk factors for dementia and research shows physical activity can delay cognitive decline in older adults. However, much of this research has used subjective physical activity data and a single follow‐up cognitive assessment. Further studies using objectively measured physical activity and comprehensive cognitive data measured at multiple timepoints are required.MethodsParticipants were 199 community‐dwelling cognitively normal older adults (68.7 5.9 years) from the Australian Imaging, Biomarkers and Lifestyle (AIBL) study. Actigraphy was used to measure physical activity at baseline, yielding measures of intensity (peak counts), total activity (total counts) and energy expenditure (kilocalories; k/cal). Cognitive function was assessed using a cognitive battery administered every 18‐months from baseline (3‐11 years follow‐up), yielding composite scores for episodic memory, executive function, attention and processing speed, and global cognition.ResultsHigher baseline energy expenditure predicted improvements in episodic memory and maintained global cognition over time (β = 0.011, SE = 0.005, p = 0.031; β = 0.009, SE = 0.004, p = 0.047, respectively). Both physical activity intensity and total activity predicted global cognition, such that those with higher peak and total counts had better cognition over time (β = 0.012, SE = 0.004, p = 0.005; β = 0.012, SE = 0.004, p = 0.005, respectively). Finally, higher total activity predicted improved episodic memory over time (β = 0.011, SE = 0.005, p = .022).ConclusionThese results suggest that physical activity is associated with preserved cognitive function over time, and that activity intensity may play an important role. This research further highlights the importance of early intervention to prevent cognitive decline and may aid in informing lifestyle interventions for dementia prevention.

Transitioning to Elecsys automated immunoassay for fresh CSF biomarkers to support the diagnosis of Alzheimer’s disease in Australia

AbstractBackgroundAlzheimer’s disease (AD) biomarker proteins, amyloid β1‐42 (Aβ42) and Tau (Phospho‐tau 181 [pTau] or Total‐tau [tTau]), in the cerebrospinal fluid (CSF) are used to support the clinical diagnosis of AD, and for enrolment of patients into therapeutic trials. Currently Roche Elecsys Generation I (GI) for frozen CSF is used, which will be replaced by Generation II (GII) for routine diagnostics use. This study examined the correlation of the biomarkers obtained from the automated Elecsys GI vs the manual INNOTEST immunoassays and the concordance of diagnostics outcome between the two methods. In addition, we also examined performance of GII on fresh CSF.MethodCerebrospinal fluid (CSF) samples were collected from local hospitals/clinics and from the study cohort of the Australian Imaging Biomarkers and Lifestyle study of Aging (AIBL). Aβ42, pTau and tTau were measured with both Elecsys and INNOTEST. Most recent collected CSF with a paired frozen and fresh CSF were assayed in parallel of both GI and GII.ResultIn the cohort of 34 collected frozen CSF from the AIBL cohort, the Spearman r correlations between INNOTEST and Elecsys methods were 0.91, 0.98, and 0.97 for Aβ42, pTau181, tTau, respectively. From 166 clinical diagnostic CSF samples (from local hospitals and clinics), the correlation was 0.89, 0.93, 0.88, respectively. Using the Aβ42 and pTau as primary parameters, the diagnostic outcome is highly concordant between the two methods in these two groups. Initial data from the paired frozen and fresh CSF showed GI and GII correlated well (r = 0.9643). Assessments on the GII cohort are ongoing.ConclusionThe automated Elecsys is highly correlated with the INNOTEST and offers high throughput and excellent precision with lower interassay/interlaboratory variation. This supports the use of Elecsys for CSF AD diagnostics in routine practice. The availability of fresh CSF analysis on GII will make the test more easily accepted into routine pathology setting.

Systemic perturbations of the kynurenine pathway precede progression to dementia independently of amyloid-β

Increasing evidence suggests that kynurenine pathway (KP) dyshomeostasis may promote disease progression in dementia. Studies in Alzheimer's disease (AD) patients confirm KP dyshomeostasis in plasma and cerebrospinal fluid (CSF) which correlates with amyloid-β and tau pathology. Herein, we performed the first comprehensive study assessing baseline levels of KP metabolites in participants enrolling in the Australian Imaging Biomarkers Flagship Study of Aging. Our purpose was to test the hypothesis that changes in KP metabolites may be biomarkers of dementia processes that are largely silent. We used a cross-sectional analytical approach to assess non-progressors (N = 73); cognitively normal (CN) or mild cognitive impairment (MCI) participants at baseline and throughout the study, and progressors (N = 166); CN or MCI at baseline but progressing to either MCI or AD during the study. Significant KP changes in progressors included increased 3-hydroxyanthranilic acid (3-HAA) and 3-hydroxyanthranilic acid/anthranilic acid (3-HAA/AA) ratio, the latter having the largest effect on the odds of an individual being a progressor (OR 35.3; 95% CI between 14 and 104). 3-HAA levels were hence surprisingly bi-phasic, high in progressors but low in non-progressors or participants who had already transitioned to MCI or dementia. This is a new, unexpected and interesting result, as most studies of the KP in neurodegenerative disease show reduced 3-HAA/AA ratio after diagnosis. The neuroprotective metabolite picolinic acid was also significantly decreased while the neurotoxic metabolite 3-hydroxykynurenine increased in progressors. These results were significant even after adjustment for confounders. Considering the magnitude of the OR to predict change in cognition, it is important that these findings are replicated in other populations. Independent validation of our findings may confirm the utility of 3-HAA/AA ratio to predict change in cognition leading to dementia in clinical settings.

Non-standard pipeline without MRI has replicability in computation of Centiloid scale values for PiB and 18F-labeled amyloid PET tracers

The magnetic resonance imaging (MRI)-less non-standard pipeline for amyloid positron emission tomography (PET) published by Bourgeat et al., in 2018 calculates Centiloid scale values that are highly consistent with those computed using the standard pipeline. The purpose of this study was to demonstrate that the non-standard pipeline can compute Centiloid scale values in high agreement with the standard pipeline when using different datasets of amyloid PET tracers in our local computer environment.PET images of 11C-Pittsburgh compound B (11C-PiB), 18F-florbetapir, 18F-flutemetamol, 18F-florbetaben, and 18F-NAV4694 from the calibration dataset were processed using both the standard and non-standard pipelines, and the computed cortical standardized uptake value ratio (SUVr) value was converted to the Centiloid scale value using the method described by Klunk et al., in 2015. The conversion equations from the SUVr to Centiloid scale values for each tracer were obtained during this process. Using these equations, we compared the Centiloid scale values obtained using the standard and non-standard pipelines using the validation datasets of each tracer from the Japanese Alzheimer’s Disease Neuroimaging Initiative (J-ADNI), Alzheimer’s Disease Neuroimaging Initiative (ADNI), and Australian Imaging Biomarkers and Lifestyle (AIBL).In the calibration datasets, there was high agreement (R2 > 0.97) and slight bias between the Centiloid scale values calculated by the non-standard and standard pipelines for all tracers. Despite relatively little NAV4694 data in the validation datasets, there was high agreement between the Centiloid scale values calculated using the non-standard and standard pipelines for all tracers. The bias for florbetaben and NAV4694 using the non-standard pipeline was 1.6% underestimation and 3.3% overestimation, respectively; these values were smaller than those reported by Bourgeat et al. Analysis of outliers also suggested that the non-standard pipeline might be vulnerable to anatomical anomalies. Given the slight variance of the Centiloid scale in young controls, flutemetamol and NAV4694 might be suitable tracers for the non-standard pipelines.This study demonstrates the replicability of the non-standard pipelines across computing environments, datasets, scanners, and tracers. When MRI is not available, the non-standard pipeline may provide information to aid in visual assessment of amyloid PET.