Temporal dysregulation of PPARG-PRKAG2 co-expression in gray matter: Implications for cognitive decline and intervention targets in type 2 diabetes

Temporal dysregulation of PPARG-PRKAG2 co-expression in gray matter: Implications for cognitive decline and intervention targets in type 2 diabetes | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Temporal dysregulation of PPARG-PRKAG2 co-expression in gray matter: Implications for cognitive decline and intervention targets in type 2 diabetes Shelli R. Kesler, Kimberly A. Lewis, Heather Cuevas, Elena Flowers This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7907793/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Background: Type 2 diabetes mellitus (T2DM) is associated with increased risk for cognitive decline and diagnosis of Alzheimer’s disease (AD). The mechanisms of T2DM related dementia remain unclear. Methods: Brain magnetic resonance imaging was retrospectively obtained for 1,802 adults (age 66 +/- 9 years, 47% male) of whom N = 271 had T2DM. We applied an accelerated longitudinal design and imaging transcriptomics to non-invasively examine the group-level trajectories of PPARG and PRKAG2 co-expression in gray matter. Results: Gene expression trajectories differed significantly between T2DM and controls (χ² = 13.82, p = 0.001). Co-expression was higher in early stages and then weakened in later stages among T2DM, while remaining stable over time in controls. PPARG and PRKAG2 co-expression was significantly associated with cognitive function in controls (F = 3.17, p < 0.001) but not T2DM (F = 7.72, p = 0.299) suggesting dysregulated or failed compensatory mechanisms. Individuals with T2DM not taking metformin demonstrated unstable gene co-expression over time compared to those taking metformin (χ² = 12.42, p = 0.006). Conclusions: The convergence of PPARG -mediated metabolic remodeling and PRKAG2 /AMPK-driven energy sensing may act as a coordinated neuroprotective mechanism, upregulated in response to cellular stress in both pathological (T2DM) and normative (aging) contexts. However, these processes appear to become dysregulated in T2DM, potentially resulting in cognitive decline and increased risk for dementia. type 2 diabetes cognitive decline imaging transcriptomics MRI clinical dementia rating miRNA-20b metformin Figures Figure 1 Figure 2 Figure 3 Background Individuals with type 2 diabetes (T2DM) are at significantly increased risk for neurodegeneration and associated cognitive decline, including higher incidence of Alzheimer’s disease (AD). Potential mechanisms for T2DM-related neurodegeneration include metabolic dysfunction, neuroinflammation, cerebrovascular disease, disrupted gut microbiome, and insulin resistance ( 1 , 2 ). T2DM and AD also share several common molecular risk factors involved in metabolism, insulin signaling, inflammation, and toxin clearance ( 3 – 5 ). Many of the genes that regulate these processes are also critical for neuronal development and health. Using non-invasive imaging transcriptomics, we previously demonstrated that genes involved in insulin signaling, glucose metabolism ( IRS1, AKT1, PPARG, PRKAG2 ) and neurotransmission ( GRIN2B ) were significantly expressed in regions of T2DM-related gray matter atrophy ( 6 ). Gene expression patterns interacted in both T2DM and non-diabetic control groups to predict cognitive function. However, these patterns were significantly altered in T2DM, with PPARG (peroxisome proliferator-activated receptor gamma) taking over normal directional pathways from other genes to influence cognitive function. Specifically, higher PPARG was associated with higher PRKAG2 (protein kinase AMP-activated non-catalytic subunit gamma 2) expression which in turn was associated with lower Clinical Dementia Rating (CDR) in T2DM (i.e., +PPARG ◊ +PRKAG2 ◊ -CDR). Given that this was a significant difference in gene network structure compared to controls, increased PPARG influence may represent a compensatory mechanism in the context of T2DM neuropathology. The alteration in gene pathways occurred despite relatively preserved neuropsychological function, equivalent amyloid burden compared to controls, and lower apolipoprotein allele 4 ( APOE4 ) risk. Thus, these molecular changes may represent an early compensatory change in T2DM ( 6 ). However, our prior study was cross-sectional, and although we utilized Bayesian network analysis which determines a directional pathway model analogous to structural equation modeling, our findings were fundamentally associative. We therefore conducted a follow-up study that applied an accelerated longitudinal design in a larger sample. This approach allows for modeling temporal trajectories by combining cross-sectional data from different cohorts, each observed at a different point in time. Although each participant was observed only once, time was treated as a continuous variable, and individuals were evaluated across timepoints to approximate a longitudinal pattern. This method provides insight into group-level temporal effects, but not individual change. Thus, this represents a proof-of-concept study to support further research. We again utilized imaging transcriptomics to non-invasively evaluate gene expression patterns associated with T2DM neuropathology. Neuroimaging data is spatially co-localized with gene expression patterns from transcriptomic brain atlases to facilitate the study of in vivo molecular mechanisms underlying brain health and disease ( 7 – 10 ). By linking gene expression patterns directly to brain structure and function, imaging transcriptomics opens new avenues for exploring the molecular mechanisms of neurological diseases and for developing targeted therapies. Several imaging transcriptomics studies have been conducted in neurodegenerative diseases ( 11 – 13 ) but very few have involved T2DM ( 14 – 17 ). Based on our prior findings, we hypothesized that co-expression of PPARG and PRKAG2 would change over time in T2DM compared to controls. We expected that T2DM would display early compensatory activation followed by later failure consistent with prior evidence regarding neurologic compensation in neurodegeneration ( 18 ). Methods Participants We obtained data from the Mayo Clinic Study of Aging (MCSA) via the Laboratory of Neuroimaging’s Image and Data Archive at the University of Southern California. The MCSA is a prospective, epidemiologic study of cognitive aging that began in 2004 in Olmsted County, Minnesota ( 19 ). The Mayo Clinic and Olmsted Medical Center institutional review boards approved the parent study from which this data was extracted and gave approval for the sharing of anonymized data. All participants provided written informed consent during the parent study. For the current study, privacy rights of participants were observed, and no personally identifiable information was obtained. The MCSA is a longitudinal study, but only cross-sectional data is currently available in the Image and Data Archive. However, participants in the cross-sectional cohort were assessed at different time points (i.e., no repeated measures; cross-sectional data at multiple timepoints from different individuals). Thus, these data lend themselves well to an accelerated longitudinal design for approximating a longitudinal trajectory across individuals. Visits were separated by approximately 1 year across 10 years. Table 1 indicates the sample sizes at each Visit. Given that visits 10 and 11 had only one participant’s data available, these visits were excluded from the analyses. During the MCSA, participants underwent a medical history review, neurological evaluation, and provided demographic information. Trained nurses reviewed medical records to obtain data on chronic conditions including diagnosis of T2DM ( 20 , 21 ). At the time of our study, there were N = 22,239 cases in the Image and Data Archive, but only 1,802 had magnetic resonance imaging (MRI) data and N = 271 of these had T2DM. During the MCSA, participants were administered several neurological and neuropsychological assessments. For this study, we examined data from the CDR given our prior results ( 6 ). Additionally, participants underwent genotyping to determine APOE4 status (positive or negative). Gray Matter Volume Measurement. Anatomic brain MRIs were performed on all consenting individuals with no contraindications. Those who underwent MRI were not significantly different demographically from those who did not, per previous report ( 22 ). Imaging data were collected with a 3 Tesla GE scanner (GE Medical Systems, Milwaukee, WI) using a 3D magnetization prepared rapid acquisition gradient echo sequence (repetition time = 2300 ms, echo time = 3 ms, inversion time = 900 ms, flip angle = 8°, field of view = 26 cm, in-plane matrix = 256 × 256, phase field of view = 0.94, slice thickness = 1.2 mm). We measured gray matter volumes from anatomic brain MRI using voxel-based morphometry implemented in Statistical Parametric Mapping (SPM) 12 (Wellcome Trust Centre for Neuroimaging, London, UK) ( 23 ). Successful segmentation and normalization were confirmed using visual and quantitative quality assurance methods. We restricted gray matter measurement to regions of atrophy that were observed in our prior study. Imaging Transcriptomics . We obtained brain transcriptome data from the Allen Human Brain Atlas (AHBA), which is currently considered the most comprehensive transcriptional brain map available ( 7 ). AHBA was developed using six adult human donor brains to provide expression data from tens of thousands of genes measured from thousands of brain regions ( 24 ). To determine the co-localization of brain imaging and transcription data, we used the Multimodal Environment for Neuroimaging and Genomic Analysis toolbox v3.1 ( 25 ) to correlate gray matter volume with patterns of AHBA expression from the PPARG and PRKAG2 . Principal Components Analysis was performed on a region-by-gene expression matrix to identify components explaining at least 95% variance across the AHBA donors. The component scores were then entered as independent variables into a multiple linear regression analysis with the corresponding gray matter volume values as the dependent variable. The accuracy of the probe to gene mapping was verified and probe data were normalized to z-scores. Representative probes were then selected in a data-driven manner considering between-donor homogeneity and the distributions of probe data. Genomic autocorrelations were calculated using an intraclass correlation coefficient (ICC) to measure the gene expression reliability across donors. Statistical Analyses. Demographics were compared between groups at each visit using ANOVA or Chi squared tests, as appropriate. R squared values for PPARG and PRKAG2 were extracted for each participant. These values represent the strength of the image-transcriptome co-localization. As shown in Table 1, data density decreased substantially after visit 5, but restricting analyses to visits 1–4 would have excluded later stages of the trajectory and biased inferences toward earlier phases. Thus, to evaluate the co-expression trajectory of PPARG and PRKAG2 across study visits, we used generalized additive models (GAMs), which are well-suited for irregular or sparse longitudinal data ( 26 – 28 ). GAMs stabilize estimates at timepoints with limited observations by borrowing strength from neighboring values through smooth basis functions, while still allowing for localized deviations where sufficient data exist. To prevent overfitting, particularly in sparsely sampled later visits, we used a basis dimension parameter (k = 5), which controls the complexity of the smooth and acts as a regularization term. We evaluated whether the relationship between visit and PPARG expression in predicting PRKAG2 differed between groups by comparing two GAMs: one with a shared smooth surface for visit across all participants (joint-smooth model) and another with separate smooths stratified by T2DM group (stratified-smooth model). Both models included fixed effects for T2DM status, a smooth term for age, and linear terms for biological sex and body mass index (BMI). Models were compared using an approximate likelihood ratio test, given that GAMs use penalized likelihood. We additionally reported the change in Akaike Information Criterion (AIC) to aid interpretation. We conducted a supplemental analysis using a traditional linear regression model to further support the GAM findings given the sparse data in later visits. We divided the cohort into visit stages with “early” stage defined as visits 1–2 (1039 controls, 175 T2DM) and “late” stage (396 controls, 96 T2DM) as Visits 3 and higher. This division was based on the results from the GAM (see below for details), which indicated that in T2DM, the strongest co-expression between PPARG and PRKAG2 occurred at Visits 1 and 2. At Visit 3, the marginal effect of PPARG on PRKAG2 weakened. The linear regression modeled PRKAG2 expression from the interaction between PPARG , visit stage, and T2DM status (i.e., PPARG x visit stage x T2DM) with age, biological sex, and BMI as additional covariates. To examine whether the co-expression of PPARG and PRKAG2 was associated with cognitive function and whether this relationship differed by T2DM status, we fitted a GAM with CDR sum as the dependent variable. The model included a group-stratified bivariate smooth interaction between PPARG and PRKAG2 expression, allowing the gene-gene association to vary flexibly between T2DM and controls. Covariates included T2DM group, a smooth term for visit to capture nonlinear temporal trends, a smooth term for age, and linear terms for biological sex and BMI. Finally, we compared PRKAG2 expression in the T2DM group between those who were or were not taking metformin. The GAM included PPARG , metformin (yes/no), age, sex, and BMI. Data analyses were conducted in the R Statistical Package (v4.5.1, R Foundation, Vienna, Austria). GAM models were conducted using the “mgcv” library with restricted maximum likelihood for primary models and maximum likelihood for model comparisons. Alpha level for all analyses was p < 0.05. Results Mean age across groups was 65.94 +/- 8.55 years at visit 1 and 85.37 +/-2.3 years at visit 9 representing a 20-year age span across approximately 9 years of assessments (Table 1). There were no significant group-by-visit effects in terms of age (F = 1.001, p = 0.428), education (F = 1.45, p = 0.169), sex (χ² = 10.46, p = 0.234), or APOE4 status (χ² = 3.18, p = 0.074). BMI was higher in the T2DM group during earlier visits (F = 2.56, p = 0.009). PRKAG2 expression changed significantly over time for T2DM (F = 69.99, p < 0.001) and controls (F = 270.23, p < 0.001). The difference in trajectories was significant (χ² = 13.82, p = 0.001, Δ AIC = -9.93). Male sex (t = 2.49, p = 0.013) and smoothed age (F = 16.71, p < 0.001) were significant contributors to the model. In T2DM, PPARG strongly predicted higher PRKAG2 in early stages. This effect weakened or plateaued in later visits. In controls, the relationship between PPARG and PRKAG2 was stable over time (Fig. 1 ). The supplemental linear regression model of early vs. late stage supported the GAM results. The PPARG x T2DM interaction was significant and positive (estimate = 0.121, p = 0.002) indicating that PPARG predicts higher PRKAG2 in T2DM in the early stage. The PPARG x stage x T2DM interaction was significant and negative (estimate = -0.140, p = 0.039) suggesting that the relationship between PPARG and PRKAG2 declines (or reverses) in the late stage. The stage x T2DM interaction was significant and positive (estimate = 0.041, p = 0.048) indicating that individuals with T2DM in the late stage have higher baseline PRKAG2 , controlling for PPARG . PPARG - PRKAG2 co-expression significantly influenced CDR in controls (F = 3.17, p < 0.001) but was not significant for T2DM (F = 7.72, p = 0.299). Visualization of the model (Fig. 2 ) indicated a non-monotonic (inverted U-shaped) relationship between gene co-expression and CDR in T2DM. Alternatively, controls demonstrated a more stable, linear association with CDR. Age was a significant contributor to the model (F = 16.70, p 0.203). Individuals with T2DM who reported taking metformin demonstrated significantly higher PRKAG2 expression across visits compared to those who were not taking metformin (t = 2.39, p = 0.018). Expression changed over time in the metformin positive (F = 39.02, p < 0.001) and metformin negative (F = 32.45, p < 0.001) groups and these trajectories were significantly different (χ² = 12.42, p = 0.006, Δ AIC = -4.62). As shown in Fig. 3 , gene co-expression was stable and positive for the metformin group but was less stable, fluctuating over time, and showing more variability in direction and magnitude in the metformin negative group. Discussion We previously observed significant alteration in the brain transcriptome of individuals with T2DM compared to controls with a potential compensatory associative pathway from PPARG to PRKAG2 supporting cognitive function ( 6 ). In the present study, we aimed to further evaluate this relationship by examining the pseudo-longitudinal trajectory of PPARG-PRKAG2 co-expression in gray matter tissue and the effect of this trajectory on cognitive function. In T2DM, PPARG strongly predicted higher PRKAG2 in early stages, consistent with a compensatory mechanism. This effect weakened or plateaued in later visits, possibly due to loss of regulatory control or biological exhaustion. In controls, the relationship between PPARG and PRKAG2 was stable over time, likely reflecting normal regulatory function without disease-driven compensation or failure. This finding was supported by supplemental stage-based modeling which indicated that, in early-stage T2DM, higher PPARG was associated with higher PRKAG2 expression, consistent with compensatory upregulation. In late-stage T2DM, this compensatory relationship reversed, possibly reflecting system failure or overload. These findings are consistent with prior evidence regarding functional compensation in the context of neuropathology. The brain is often capable of reorganization to counteract decline and preserve function. However, ongoing pathology tends to overwhelm these compensatory mechanisms, resulting in functional decline ( 18 ). PPARG encodes peroxisome proliferator-activated receptor gamma, a transcription factor central to lipid metabolism, adipogenesis, and insulin sensitivity ( 29 ). PPARG activation downregulates NF-κB signaling, reducing neuroinflammation, which is a hallmark of both aging and T2DM-associated cognitive decline ( 30 , 31 ). Additionally, upregulation of PPARG may counteract oxidative stress and insulin resistance in the brain ( 32 ). PRKAG2 is a regulatory subunit of AMP-activated protein kinase (AMPK), a master energy sensor that restores energy balance by stimulating catabolic pathways and inhibiting anabolic processes ( 4 ). AMPK activation is critical in maintaining ATP levels during metabolic stress, which is common in both aging brains and T2DM ( 33 ). PRKAG2 is also involved in the clearance of damaged proteins and toxins, which accumulate during neurodegeneration. Accordingly, PRKAG2 expression has been shown to be elevated in AD and is associated with cognitive impairment in T2DM ( 3 , 34 ). With aging or disease, neurons experience reduced mitochondrial efficiency, increased oxidative stress, and altered glucose metabolism. A rise in PPARG–PRKAG2 co-expression may represent a compensatory adaptation with PPARG driving transcriptional programs that enhance lipid utilization, and AMPK signaling helping to restore energy balance. Their co-expression could synergistically buffer against metabolic decline in the aging or T2DM brain. We showed that both T2DM and control groups demonstrated increased PRKAG2 expression over time consistent with normal age-related changes. However, T2DM showed significantly higher PRKAG2 expression in later stages despite de-coupling with PPARG , potentially reflecting dysregulation of compensatory processes or accelerated aging. Prior studies have demonstrated evidence of accelerated cortical brain aging in individuals with T2DM ( 35 , 36 ) but our findings represent the first evidence of accelerated molecular brain aging. PPARG agonists, especially pioglitazone, show protective actions in experimental models of AD and diabetic encephalopathy ( 37 ). In rodent studies, pioglitazone reversed beta-amyloid–induced memory deficits and restored learning functions ( 38 , 39 ). In human trials involving patients with mild AD or amnestic mild cognitive impairment, pioglitazone improved cognitive scores ( 40 , 41 ). PPARG agonists work in part by reducing amyloid-beta levels, lowering inflammatory and oxidative markers, stabilizing brain-derived neurotrophic factor, and improving measures of insulin sensitivity ( 42 , 43 ). However, our findings indicate that upregulation of PPARG may shift from adaptive to maladaptive over time in the T2DM brain. Prior research has shown that excessive or prolonged PPARG activation is linked to metabolic dysregulation such as overeating and weight gain through neuronal pathways regulating energy balance ( 44 , 45 ). In our cohort, participants with T2DM initially met criteria for obesity (BMI > 30), but later declined below this threshold, which may reflect early PPARG upregulation followed by later downregulation or improved diabetes management. Yet BMI did not contribute significantly to our gene expression or cognitive models, suggesting that PPARG- related brain effects may operate independently from adiposity. Because BMI is a coarse measure of metabolic status, factors such as insulin resistance, adipose distribution, inflammation, or diet quality may better reflect PPARG -related metabolic variance. Incorporating these refined measures into future studies could clarify how peripheral metabolic remodeling and central PPARG signaling interact over time to influence neurocognitive trajectories of T2DM. Further, we did not have access to blood assay data to examine the potential effects of blood glucose level or inflammatory markers on our findings. However, this research emphasizes the need for a balanced homeostatic regulation between neuroprotective and metabolic pathways to sustain cognitive function in T2DM and highlight the importance of metabolic phenotyping beyond BMI. MicroRNAs regulate mRNA translation and can cross the blood brain barrier, creating the possibility that peripherally administered therapeutic interventions might influence central nervous system function. Our prior study showed that circulating miR-20b is inversely associated with longitudinal changes in fasting glucose ( 46 ) and PPARG is a functionally validated mRNA target of miR-20b ( 47 ). We also showed that circulating miR-20b decreased after two years of treatment with metformin compared to an increase in expression in a placebo group, who had higher rates of incident T2DM ( 48 ). Because miR-20b is linked to PPARG suppression, this pattern implies that higher miR-20b expression in untreated individuals may reduce PPARG expression, theoretically contributing to the increased diabetes risk observed in the placebo group. These observations provide preliminary evidence that microRNAs such as miR-20b may be a therapeutic approach to modulate PPARG- related transcriptomic changes associated with neurodegeneration and cognitive decline in T2DM. PPARG – PRKAG2 co-expression was significantly associated with cognitive function as measured by the CDR in controls but not in T2DM. This dissociation may imply a loss of regulatory control, breakdown of compensation, or uncoupling of biological mechanisms in T2DM. The relationship in T2DM was U-shaped where higher CDR (worse cognition) occurred at both low and high gene co-expression levels, consistent with failed compensation at extremes. These findings may also suggest that alternative neurobiologic processes are recruited to support cognitive function in T2DM. Previous studies suggest that cognitive performance is preserved in T2DM by an increase in functional brain connectivity ( 49 , 50 ). Functional neuroimaging was not available for this cohort in the Imaging Data Archive and therefore, future studies that include multimodal imaging transcriptomics are required. Metformin treatment was associated with increased, stable PRKAG2 expression across visits in the T2DM group. This suggests a sustained engagement of the brain’s AMPK pathway that may contribute to metformin’s reported neuroprotective effects on neuronal metabolic equilibrium during chronic energetic stress ( 51 , 52 ). This stability suggests that metformin may recalibrate neuronal metabolism toward cognitive resilience, promoting continuous energy balance and proteostatic clearance through mTOR inhibition and autophagy induction ( 52 – 54 ). These findings align with emerging evidence that genetic variability in PRKAG2 influences individual responses to metformin ( 55 ) and that metformin helps maintain blood-brain barrier integrity via AMPK activation ( 56 ). These observations extend the role of metformin beyond glycemic control, positioning it as a modulator of brain bioenergetic homeostasis that could delay or attenuate the molecular cascades linked to T2DM, cognitive decline, and AD ( 57 ). Future mechanistic and translational work should determine whether persistent PRKAG2 stabilization represents a biomarker of adaptive neural metabolism or a viable target for interventions aimed at preventing diabetes-related neurodegeneration. There are several limitations to this study. We utilized an accelerated longitudinal design which provides insight into group-level trajectories but not individual changes. True longitudinal studies are needed to evaluate change in brain transcriptome pathways among individuals with T2DM and the impact of these changes on cognitive function and risk for AD. We examined the co-expression of PPARG and PRKAG2 because this was the primary altered pathway related to cognitive function in our initial study of T2DM and to reduce comparisons considering the sample size. However, alteration of other brain transcriptome pathways may also be important for cognitive function in T2DM. Data in later visits were very sparse, and we addressed this using penalized GAMs and a supplementary analysis of early versus late visits. However, analyses may have lacked power to detect alternative trajectories. While the AHBA is the most comprehensive brain transcriptome atlas to date, it was developed based on a small sample. The atlas includes gene expression patterns for thousands of brain regions but does not comprehensively cover the entire brain. In conclusion, our findings suggest that alterations of PPARG-PRKAG2 co-expression may be involved in the brain’s response to both normative aging and neuropathological conditions associated with T2DM. However, this response appears to become dysregulated over time in T2DM. Our results provide novel insights regarding the potential molecular mechanisms of T2DM-related cognitive decline and the risk for AD. Further research is required that includes younger adults with T2DM to determine how early these changes occur. Data regarding blood glucose levels would also provide important insights regarding the impact of T2DM disease management on the brain transcriptome. Multimodal neuroimaging will be necessary to further examine functional compensatory changes and to investigate the role of the brain transcriptome in these changes. Abbreviations T2DM type 2 diabetes AD Alzheimer’s disease PPARG Peroxisome Proliferator-Activated Receptor Gamma PRKAG2 Protein Kinase AMP-Activated Non-Catalytic Subunit Gamma 2 MRI Magnetic resonance imaging AHBA Allen Human Brain Atlas GAM Generalized additive model CDR Clinical Dementia Rating BMI Body mass index AIC Akaike Information Criterion MCSA Mayo Clinic Study of Aging AMPK AMP-activated protein kinase Declarations Ethics Approval and Consent to Participate The Mayo Clinic and Olmsted Medical Center institutional review boards approved the parent study from which this data was extracted and gave approval for the sharing of anonymized data. All participants provided written informed consent during the parent study. Consent for Publication Not applicable. Competing Interests The authors declare that they have no competing interests. Funding The data contained in this analysis were obtained under one of the following research grants from the National Institutes of Health to the Mayo Clinic Study of Aging (U01 AG06786, Ronald Petersen, PI) or the Mayo Alzheimer’s Disease Research Center (P30 AG062677, Ronald Petersen, PI). However, the authors of this manuscript did not receive funding for this study. Author Contribution SRK: Conceptualization, Data curation, Formal analysis, Methodology, Project administration, Supervision, Validation, Writing - original draft, Writing - review & editing. KAL: Validation, Methodology, Writing - original draft, Writing - review & editing. HC: Methodology, Validation, Writing - original draft, Writing - review & editing. EF: Methodology, Validation, Writing - original draft, Writing - review & editing. Data Availability The dataset supporting the conclusions of this article is available to qualified researchers and industry partners from the Laboratory of Neuroimaging’s Image and Data Archive at the University of Southern California (http://loni.usc.edu) or the Global Alzheimer’s Association Interactive Network (https://www.gaain.org). References Blazquez E, Velazquez E, Hurtado-Carneiro V, Ruiz-Albusac JM. Insulin in the brain: its pathophysiological implications for States related with central insulin resistance, type 2 diabetes and Alzheimer's disease. Front Endocrinol (Lausanne). 2014;5:161. Lazar E, Sherzai A, Adeghate J, Sherzai D. Gut dysbiosis, insulin resistance and Alzheimer's disease: review of a novel approach to neurodegeneration. Front Biosci (Schol Ed). 2021;13(1):17–29. Bharadwaj P, Martins RN. PRKAG2 gene expression is elevated and its protein levels are associated with increased amyloid-β accumulation in the Alzheimer’s disease brain. J Alzheimer’s Disease. 2020;74(2):441–8. Randrianarisoa E, Lehn-Stefan A, Krier J, Böhm A, Heni M, De Hrabě M, et al. AMPK Subunits Harbor Largely Nonoverlapping Genetic Determinants for Body Fat Mass, Glucose Metabolism, and Cholesterol Metabolism. J Clin Endocrinol Metabolism. 2020;105(1):14–25. Shah K, DeSilva S, Abbruscato T. The role of glucose transporters in brain disease: diabetes and Alzheimer’s disease. Int J Mol Sci. 2012;13(10):12629–55. Kesler SR, Cuevas H, Lewis KA, Franco-Rocha OY, Flowers E. The expression of insulin signaling and N-methyl-D-aspartate receptor genes in areas of gray matter atrophy is associated with cognitive function in type 2 diabetes. J Alzheimers Dis. 2025:13872877251364906. Arnatkeviciute A, Fulcher BD, Bellgrove MA, Fornito A. Imaging Transcriptomics of Brain Disorders. Biol Psychiatry Glob Open Sci. 2022;2(4):319–31. Martins D, Giacomel A, Williams SCR, Turkheimer F, Dipasquale O, Veronese M, et al. Imaging transcriptomics: Convergent cellular, transcriptomic, and molecular neuroimaging signatures in the healthy adult human brain. Cell Rep. 2021;37(13):110173. Sunkin SM, Ng L, Lau C, Dolbeare T, Gilbert TL, Thompson CL, et al. Allen Brain Atlas: an integrated spatio-temporal portal for exploring the central nervous system. Nucleic Acids Res. 2013;41(Database issue):D996–1008. Grothe MJ, Sepulcre J, Gonzalez-Escamilla G, Jelistratova I, Scholl M, Hansson O, et al. Molecular properties underlying regional vulnerability to Alzheimer's disease pathology. Brain. 2018;141(9):2755–71. Lancour D, Dupuis J, Mayeux R, Haines JL, Pericak-Vance MA, Schellenberg GC, et al. Analysis of brain region-specific co-expression networks reveals clustering of established and novel genes associated with Alzheimer disease. Alzheimers Res Ther. 2020;12(1):1–11. Baik JY, Kim M, Bao J, Long Q, Shen L, Initiative ADN. Identifying Alzheimer’s genes via brain transcriptome mapping. BMC Med Genom. 2022;15(Suppl 2):116. Mroczek M, Desouky A, Sirry W. Imaging transcriptomics in neurodegenerative diseases. J Neuroimaging. 2021;31(2):244–50. Liu Q, Du X, Zhang Y, Ding H, Qin W, Zhang Q. Associations between morphometric similarity network and brain gene expression in type 2 diabetes mellitus. Neuroscience. 2025;572:84–92. Zhang Y, Du X, Qin W, Fu Y, Wang Z, Zhang Q. Association between gene expression and altered resting-state functional networks in type 2 diabetes. Front Aging Neurosci. 2023;15:1290231. Mullins RJ, Diehl TC, Chia CW, Kapogiannis D. Insulin Resistance as a Link between Amyloid-Beta and Tau Pathologies in Alzheimer's Disease. Front Aging Neurosci. 2017;9:118. Huang X, He YX, Wan S. Genetic mechanisms of hemispheric functional connectivity in diabetic retinopathy: a joint neuroimaging and transcriptomic study. Front Cell Dev Biol. 2025;13:1590627. Scheller E, Minkova L, Leitner M, Kloppel S. Attempted and successful compensation in preclinical and early manifest neurodegeneration - a review of task FMRI studies. Front Psychiatry. 2014;5:132. Roberts RO, Geda YE, Knopman DS, Cha RH, Pankratz VS, Boeve BF, et al. The Mayo Clinic Study of Aging: design and sampling, participation, baseline measures and sample characteristics. Neuroepidemiology. 2008;30(1):58–69. Wennberg AMV, Hagen CE, Edwards K, Roberts RO, Machulda MM, Knopman DS, et al. Association of antidiabetic medication use, cognitive decline, and risk of cognitive impairment in older people with type 2 diabetes: Results from the population-based Mayo Clinic Study of Aging. Int J Geriatr Psychiatry. 2018;33(8):1114–20. Alhurani RE, Vassilaki M, Aakre JA, Mielke MM, Kremers WK, Machulda MM, et al. Decline in Weight and Incident Mild Cognitive Impairment: Mayo Clinic Study of Aging. JAMA Neurol. 2016;73(4):439–46. Whitwell JL, Wiste HJ, Weigand SD, Rocca WA, Knopman DS, Roberts RO, et al. Comparison of imaging biomarkers in the Alzheimer Disease Neuroimaging Initiative and the Mayo Clinic Study of Aging. Arch Neurol. 2012;69(5):614–22. Kurth F, Luders E, Gaser C. Voxel-Based Morphometry. Brain Mapping2015. pp. 345–9. Hawrylycz MJ, Lein ES, Guillozet-Bongaarts AL, Shen EH, Ng L, Miller JA, et al. An anatomically comprehensive atlas of the adult human brain transcriptome. Nature. 2012;489(7416):391–9. Rizzo G, Veronese M, Expert P, Turkheimer FE, Bertoldo A. MENGA: A New Comprehensive Tool for the Integration of Neuroimaging Data and the Allen Human Brain Transcriptome Atlas. PLoS ONE. 2016;11(2):e0148744. Mundo AI, Tipton JR, Muldoon TJ. Generalized additive models to analyze nonlinear trends in biomedical longitudinal data using R: Beyond repeated measures ANOVA and linear mixed models. Stat Med. 2022;41(21):4266–83. Hastie T, Tibshirani R. Generalized additive models. Stat Sci. 1986;1(3):297–310. Yilema SA, Shiferaw YA, Belay AT, Belay DB. Mapping the spatial disparities of HIV prevalence in Ethiopian zones using the generalized additive model. Sci Rep. 2024;14(1):6215. Stump M, Guo D-F, Lu K-T, Mukohda M, Cassell MD, Norris AW, et al. Nervous System Expression of PPARγ and Mutant PPARγ Has Profound Effects on Metabolic Regulation and Brain Development. Endocrinology. 2016;157(11):4266–75. Zhang Y, Chen C, Jiang Y, Wang S, Wu X, Wang K. PPARγ coactivator-1α (PGC-1α) protects neuroblastoma cells against amyloid-beta (Aβ) induced cell death and neuroinflammation via NF-κB pathway. BMC Neurosci. 2017;18(1):69. Rom S, Zuluaga-Ramirez V, Gajghate S, Seliga A, Winfield M, Heldt NA, et al. Hyperglycemia-driven neuroinflammation compromises BBB leading to memory loss in both diabetes mellitus (DM) type 1 and type 2 mouse models. Mol Neurobiol. 2019;56(3):1883–96. Wang P, Li B, Cai G, Huang M, Jiang L, Pu J et al. Activation of PPAR-γ by Pioglitazone Attenuates Oxidative Stress in Aging Rat Cerebral Arteries Through Upregulating UCP2. J Cardiovasc Pharmacol. 2014;64(6). Bonomini F, Rodella LF, Rezzani R. Metabolic syndrome, aging and involvement of oxidative stress. Aging disease. 2015;6(2):109. Kim E, Lee SH, Lee KS, Cheong H-K, Namkoong K, Hong CH, et al. AMPK γ2 subunit gene PRKAG2 polymorphism associated with cognitive impairment as well as diabetes in old age. Psychoneuroendocrinology. 2012;37(3):358–65. Antal B, McMahon LP, Sultan SF, Lithen A, Wexler DJ, Dickerson B, et al. Type 2 diabetes mellitus accelerates brain aging and cognitive decline: Complementary findings from UK Biobank and meta-analyses. eLife. 2022;11:e73138. Dove A, Wang J, Huang H, Dunk MM, Sakakibara S, Guitart-Masip M, et al. Diabetes, Prediabetes, and Brain Aging: The Role of Healthy Lifestyle. Diabetes Care. 2024;47(10):1794–802. Alhowail A, Alsikhan R, Alsaud M, Aldubayan M, Rabbani SI. Protective Effects of Pioglitazone on Cognitive Impairment and the Underlying Mechanisms: A Review of Literature. Drug Des Devel Ther. 2022;16:2919–31. Sharma B, Singh N. Behavioral and biochemical investigations to explore pharmacological potential of PPAR-gamma agonists in vascular dementia of diabetic rats. Pharmacol Biochem Behav. 2011;100(2):320–9. Liu J, Wang LN, Jia JP. Peroxisome proliferator-activated receptor-gamma agonists for Alzheimer's disease and amnestic mild cognitive impairment: a systematic review and meta-analysis. Drugs Aging. 2015;32(1):57–65. Watson GS, Cholerton BA, Reger MA, Baker LD, Plymate SR, Asthana S, et al. Preserved cognition in patients with early Alzheimer disease and amnestic mild cognitive impairment during treatment with rosiglitazone: a preliminary study. Am J Geriatr Psychiatry. 2005;13(11):950–8. Muñoz-Jiménez M, Zaarkti A, García-Arnés JA, García-Casares N. Antidiabetic Drugs in Alzheimer's Disease and Mild Cognitive Impairment: A Systematic Review. Dement Geriatr Cogn Disord. 2020;49(5):423–34. Prakash A, Kumar A. Role of nuclear receptor on regulation of BDNF and neuroinflammation in hippocampus of β-amyloid animal model of Alzheimer's disease. Neurotox Res. 2014;25(4):335–47. Titus C, Hoque MT, Bendayan R. PPAR agonists for the treatment of neuroinflammatory diseases. Trends Pharmacol Sci. 2024;45(1):9–23. Diano S, Liu Z-W, Jeong JK, Dietrich MO, Ruan H-B, Kim E, et al. Peroxisome proliferation–associated control of reactive oxygen species sets melanocortin tone and feeding in diet-induced obesity. Nat Med. 2011;17(9):1121–7. Stump M, Guo DF, Lu KT, Mukohda M, Liu X, Rahmouni K, et al. Effect of selective expression of dominant-negative PPARγ in pro-opiomelanocortin neurons on the control of energy balance. Physiol Genomics. 2016;48(7):491–501. Flowers E, Allen IE, Kanaya AM, Aouizerat BE. Circulating MicroRNAs predict glycemic improvement and response to a behavioral intervention. Biomark Res. 2021;9(1):65. Cui S, Yu S, Huang HY, Lin YC, Huang Y, Zhang B, et al. miRTarBase 2025: updates to the collection of experimentally validated microRNA-target interactions. Nucleic Acids Res. 2025;53(D1):D147–56. Lewis KA, Stroebel BM, Zhang L, Aouizerat B, Mattis AN, Flowers E. MicroRNAs Associated with Metformin Treatment in the Diabetes Prevention Program. Int J Mol Sci. 2024;25(11). Wang J, Zhou S, Deng D, Chen M, Cai H, Zhang C, et al. Compensatory thalamocortical functional hyperconnectivity in type 2 Diabetes Mellitus. Brain Imaging Behav. 2022;16(6):2556–68. Zhou B, Wang X, Yang Q, Wu F, Tang L, Wang J et al. Topological Alterations of the Brain Functional Network in Type 2 Diabetes Mellitus Patients With and Without Mild Cognitive Impairment. Front Aging Neurosci. 2022;Volume 14–2022:834319. Kim HI, Han Y, Park J. AMP-Activated Protein Kinases in Health and Disease. Int J Mol Sci. 2025;26(16):8075. Hardie DG. Keeping the home fires burning: AMP-activated protein kinase. J R Soc Interface. 2018;15(138). Hinson JT, Chopra A, Lowe A, Sheng CC, Gupta RM, Kuppusamy R, et al. Integrative Analysis of PRKAG2 Cardiomyopathy iPS and Microtissue Models Identifies AMPK as a Regulator of Metabolism, Survival, and Fibrosis. Cell Rep. 2016;17(12):3292–304. Wu S, Zou M-HAMPK. Mitochondrial Function, and Cardiovascular Disease. Int J Mol Sci [Internet]. 2020; 21(14). Xiao D, Liu J-Y, Zhang S-M, Liu R-R, Yin J-Y, Han X-Y et al. A two-stage study identifies two novel polymorphisms in PRKAG2 affecting metformin response in Chinese type 2 diabetes patients. Pharmacogenomics Personalized Med. 2021:745–55. Li N, Zhou T, Fei E. Actions of Metformin in the Brain: A New Perspective of Metformin Treatments in Related Neurological Disorders. Int J Mol Sci [Internet]. 2022; 23(15). Muñoz-Jiménez M, Zaarkti A, García-Arnés JA, García-Casares N. Antidiabetic Drugs in Alzheimer’s Disease and Mild Cognitive Impairment: A Systematic Review. Dement Geriatr Cogn Disord. 2020;49(5):423–34. Tables Table 1. Visit Group N Age Male BMI CDR Education APOE4 Metformin PPARG PRKAG2 1 Control 724 65.47 (8.4) 377 (52%) 28.37 (4.9) 0.103 (0.49) 14.88 (2.5) 216 (30%) 2 (0.28%) 0.325 (0.04) 0.294 (0.03) T2DM 121 68.70 (8.9) 70 (58%) 31.62 (5.2) 0.218 (0.74) 13.92 (2.3) 27 (22%) 73 (60%) 0.312 (0.05) 0.288 (0.04) 2 Control 315 68.44 (9.2) 152 (48%) 28.48 (5.0) 0.102 (0.41) 15.03 (2.6) 97 (31%) 1(0.32%) 0.319 (0.04) 0.289 (0.03) T2DM 54 71.51 (8.0) 33 (61%) 30.26 (3.7) 0.269 (0.70) 14.61 (2.5) 16 (30%) 32(59%) 0.312 (0.03) 0.286 (0.03) 3 Control 113 73.72 (8.6) 57 (50%) 27.22 (4.6) 0.296 (1.14) 14.92 (2.9) 34 (30%) 0(0%) 0.313 (0.04) 0.283 (0.03) T2DM 33 73.64 (8.0) 20 (61%) 33.74 (5.6) 0.273 (0.55) 13.91 (2.7) 8 (24%) 21(64%) 0.298 (0.04) 0.280 (0.03) 4 Control 155 78.98 (5.5) 78 (50%) 27.30 (4.5) 0.386 (1.10) 14.24 (3.0) 43 (28%) 0(0%) 0.306 (0.04) 0.279 (0.03) T2DM 38 79.32 (4.8) 28 (74%) 29.10 (5.3) 0.395 (1.16) 14.26 (2.6) 6 (16%) 15(40%) 0.307 (0.05) 0.275 (0.03) 5 Control 83 81.99 (4.0) 48 (58%) 26.82 (4.4) 0.470 (1.28) 14.60 (3.1) 19 (23%) 0(0%) 0.299 (0.04) 0.275 (0.03) T2DM 14 81.89 (3.5) 9 (64%) 29.89 (6.8) 0.107 (0.21) 14.57 (2.5) 5 (36%) 8(57%) 0.293 (0.03) 0.269 (0.03) 6 Control 18 81.91 (3.7) 8 (44%) 27.51 (4.9) 0.806 (1.61) 13.78 (2.6) 4 (22%) 0(0%) 0.284 (0.04) 0.273 (0.03) T2DM 5 84.16 (3.6) 3 (60%) 29.26 (4.4) 0.900 (1.25) 12.20 (3.0) 3 (60%) 1(20%) 0.299 (0.03) 0.270 (0.01) 7 Control 8 83.31 (3.6) 3 (37.5%) 27.92 (4.7) 0.188 (0.53) 14.25 (2.6) 4 (50%) 0(0%) 0.288 (0.03) 0.277 (0.03) T2DM 1 78.80 (N/A) 1 (100%) 25.77 (N/A) N/A (N/A) 16.00 (N/A) 1 (100%) 1(100%) 0.237 (N/A) 0.250 (N/A) 8 Control 14 84.74 (2.2) 10 (71%) 26.73 (3.4) 0.214 (0.38) 15.00 (3.5) 4 (29%) 0(0%) 0.314 (0.05) 0.286 (0.03) T2DM 3 85.20 (3.4) 1 (33%) 26.80 (2.1) 0.167 (0.29) 10.67 (2.3) 3 (100%) 2(67%) 0.281 (0.03) 0.282 (0.02) 9 Control 5 85.88 (2.0) 5 (100%) 26.11 (2.5) 0.100 (0.22) 16.00 (3.2) 1 (20%) 0(0%) 0.308 (0.02) 0.287 (0.02) T2DM 2 84.10 (3.4) 2 (100%) 29.32 (3.1) 0.000 (0.00) 17.00 (4.2) 0 (0%) 1(50%) 0.273 (0.08) 0.280 (0.04) Additional Declarations No competing interests reported. Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-7907793","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":532445753,"identity":"263d9b7e-b084-44b9-a679-d41a790b9a91","order_by":0,"name":"Shelli R. Kesler","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAABDklEQVRIie2PMUvDQBiGP5frcibrhWLzF95wEBCKv6UlkClgJ+lmQUiXSmfxTwiCc+QgXQ7nDoIVIVOHiEuEQI0aQaSnHR3ugeO4u/fhe4/IYvmXsGaNiNyPA76/ZL8pTdKbvIfQXuykIPsK/aX4l/nwsUL/WC60fClH94SHs1xUNfWc5WCrgrtYSY748EYnochQEHIWBecpSc+k8E7aJSiEWRI2f1GNwoPV/oSGVwbFn3WmrxU2kPO1LD8Vt7ytazo1KaRZThwZIBKIdsreE2c0gKmYZlGXI4JYFidCQ3Evj6U8SEVwoVeGYix4rsZHcOfRdTmuVc9RqvDWdd93FoZiP+HtLnaLWywWi2Urb6+JX10dJnCgAAAAAElFTkSuQmCC","orcid":"","institution":"University of Texas at Austin","correspondingAuthor":true,"prefix":"","firstName":"Shelli","middleName":"R.","lastName":"Kesler","suffix":""},{"id":532445754,"identity":"9bde5ddc-f81f-4e95-8ae9-657fe238a1b0","order_by":1,"name":"Kimberly A. Lewis","email":"","orcid":"","institution":"University of California Los Angeles","correspondingAuthor":false,"prefix":"","firstName":"Kimberly","middleName":"A.","lastName":"Lewis","suffix":""},{"id":532445755,"identity":"a641259f-9669-46a4-8445-f620bcf81985","order_by":2,"name":"Heather Cuevas","email":"","orcid":"","institution":"University of Texas at Austin","correspondingAuthor":false,"prefix":"","firstName":"Heather","middleName":"","lastName":"Cuevas","suffix":""},{"id":532445756,"identity":"f3f7dd28-49fc-4767-822a-ec415d6fba28","order_by":3,"name":"Elena Flowers","email":"","orcid":"","institution":"University of California","correspondingAuthor":false,"prefix":"","firstName":"Elena","middleName":"","lastName":"Flowers","suffix":""}],"badges":[],"createdAt":"2025-10-20 17:13:45","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7907793/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7907793/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":94136056,"identity":"a74f3143-bac5-43dc-9831-1b1062cf2c27","added_by":"auto","created_at":"2025-10-22 19:02:11","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":16523625,"visible":true,"origin":"","legend":"","description":"","filename":"PPARGMolneuro102025.docx","url":"https://assets-eu.researchsquare.com/files/rs-7907793/v1/e3600cbe33a2f3c3bedf10e5.docx"},{"id":94136044,"identity":"2e9be010-3e7d-4cb6-9348-7bd3f2acf9da","added_by":"auto","created_at":"2025-10-22 19:02:10","extension":"json","order_by":1,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":6388,"visible":true,"origin":"","legend":"","description":"","filename":"5f560ccce22f40e79b022d83a1770a18.json","url":"https://assets-eu.researchsquare.com/files/rs-7907793/v1/d4978507d9eae29cf27a8f21.json"},{"id":94136525,"identity":"26effc74-ff5a-460b-86bd-f77220234be7","added_by":"auto","created_at":"2025-10-22 19:10:10","extension":"xml","order_by":2,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":124179,"visible":true,"origin":"","legend":"","description":"","filename":"5f560ccce22f40e79b022d83a1770a181enriched.xml","url":"https://assets-eu.researchsquare.com/files/rs-7907793/v1/9d6c704abc41907380f8e317.xml"},{"id":94136043,"identity":"9fb4e564-7900-42f8-a6b6-5cebf45804b0","added_by":"auto","created_at":"2025-10-22 19:02:10","extension":"jpeg","order_by":3,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1966,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage1.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7907793/v1/bfb36e47f3e24837e0767631.jpeg"},{"id":94136524,"identity":"2984e43d-0d1a-4bce-8d55-830fab72a2c6","added_by":"auto","created_at":"2025-10-22 19:10:10","extension":"jpeg","order_by":4,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1226263,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage2.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7907793/v1/5235b4dfda39a3cc55151a9c.jpeg"},{"id":94136050,"identity":"49650f5c-4ef2-4915-b770-4577f96a02f3","added_by":"auto","created_at":"2025-10-22 19:02:11","extension":"jpeg","order_by":5,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":5464454,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage3.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7907793/v1/f026b823a75c532ba83335e7.jpeg"},{"id":94136527,"identity":"045fb0a9-15d4-4cd3-84ad-529a3a4db18f","added_by":"auto","created_at":"2025-10-22 19:10:11","extension":"jpeg","order_by":6,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":5464454,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage4.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7907793/v1/9a7e9988c6ae6bd3b6ed0ae2.jpeg"},{"id":94136052,"identity":"f17b9ea6-9e16-4dc5-9fe0-1cd2b1fce83c","added_by":"auto","created_at":"2025-10-22 19:02:11","extension":"jpeg","order_by":7,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":394925,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage5.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7907793/v1/5448ce06936af490ea98f76e.jpeg"},{"id":94136595,"identity":"ff4a1305-903b-4628-afbd-424dddf9d57f","added_by":"auto","created_at":"2025-10-22 19:18:11","extension":"png","order_by":8,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":999,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-7907793/v1/bac467f996401904e662c0ec.png"},{"id":94136057,"identity":"cb2b5f0c-ee46-47de-9840-156a208d915b","added_by":"auto","created_at":"2025-10-22 19:02:11","extension":"png","order_by":9,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":659718,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-7907793/v1/6e40ab2052daf78009fcca20.png"},{"id":94136047,"identity":"ebde14bb-77d4-4b3b-b58f-cf4bd2228b09","added_by":"auto","created_at":"2025-10-22 19:02:10","extension":"png","order_by":10,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":43092,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-7907793/v1/798cb30841a4635ca660fa2d.png"},{"id":94136045,"identity":"98e07d62-9351-4831-becf-5402e390e0dc","added_by":"auto","created_at":"2025-10-22 19:02:10","extension":"png","order_by":11,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":53725,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-7907793/v1/4aabd573e054890af9bc4520.png"},{"id":94136053,"identity":"5c634898-b804-4297-a031-608ae9861ac3","added_by":"auto","created_at":"2025-10-22 19:02:11","extension":"png","order_by":12,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":77575,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-7907793/v1/6bc9369314d0b16f2263a485.png"},{"id":94136054,"identity":"32b9e02d-ab43-44b9-a305-f81d8f89c516","added_by":"auto","created_at":"2025-10-22 19:02:11","extension":"xml","order_by":13,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":122781,"visible":true,"origin":"","legend":"","description":"","filename":"5f560ccce22f40e79b022d83a1770a181structuring.xml","url":"https://assets-eu.researchsquare.com/files/rs-7907793/v1/ed43d508aa3a91dda7d3e8b1.xml"},{"id":94136058,"identity":"43e4d0f5-aa0a-4ea5-91c7-c1273d6381aa","added_by":"auto","created_at":"2025-10-22 19:02:11","extension":"html","order_by":14,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":132631,"visible":true,"origin":"","legend":"","description":"","filename":"earlyproof.html","url":"https://assets-eu.researchsquare.com/files/rs-7907793/v1/b6a6a47c0b2ede36d4a877b9.html"},{"id":94136041,"identity":"502a7061-e9f6-4753-b8fb-729889226053","added_by":"auto","created_at":"2025-10-22 19:02:10","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":368607,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003ePRKAG2\u003c/em\u003eExpression Across Visits. T2DM and controls demonstrated significantly different trajectories of \u003cem\u003ePRKAG2\u003c/em\u003e expression over time (χ² = 13.82, p = 0.001). Top panels: In T2DM, \u003cem\u003ePPARG\u003c/em\u003ewas strongly associated with higher \u003cem\u003ePRKAG2\u003c/em\u003e expression during early visits, but this relationship diminished in later visits. In controls, the \u003cem\u003ePPARG–PRKAG2\u003c/em\u003erelationship was more stable across time. Each surface was adjusted for covariates. \u0026nbsp;Contour lines represent constant predicted values of \u003cem\u003ePRKAG2\u003c/em\u003e across combinations of visit and \u003cem\u003ePPARG\u003c/em\u003eexpression, based on the fitted GAM surface. Curved lines indicate a strong interaction effect between \u003cem\u003ePPARG\u003c/em\u003eand visit, and tighter spaced lines indicate a rapid change in \u003cem\u003ePRKAG2\u003c/em\u003e levels. Bottom panel: partial effects of smoothed covariates. The solid line indicates the estimated smooth function with the shaded area representing the 95% confidence interval.\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-7907793/v1/26c90bf041ee0901ab0375b9.png"},{"id":94136040,"identity":"6628f5ab-75d2-4719-8e3e-ca3b6cf62ad9","added_by":"auto","created_at":"2025-10-22 19:02:10","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":416900,"visible":true,"origin":"","legend":"\u003cp\u003eEffect of \u003cem\u003ePPARG-PRKAG2\u003c/em\u003eCo-expression on CDR. \u003cem\u003ePPARG\u003c/em\u003e-\u003cem\u003ePRKAG2\u003c/em\u003eco-expression was significantly associated with CDR in controls (F = 3.17, p \u0026lt; 0.001) but not in T2DM (F = 7.72, p = 0.299). Top panels: contour color gradients represent CDR values (warmer shades = higher predicted impairment). Each surface was adjusted for covariates. \u0026nbsp;The T2DM group showed a nonlinear co-expression pattern suggestive of dysregulated compensatory dynamics, whereas controls showed a more stable, linear surface. Bottom panels: partial effects of smoothed covariates. The solid line indicates the estimated smooth function with the shaded area representing the 95% confidence interval.\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-7907793/v1/8888113a0e14a346b731ed36.png"},{"id":94136523,"identity":"4f451837-4a23-45f6-b1df-08aa048496fa","added_by":"auto","created_at":"2025-10-22 19:10:10","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":350488,"visible":true,"origin":"","legend":"\u003cp\u003eDifferences in PPARG-PRKAG2 Co-expression Associated with Metformin. Individuals with T2DM taking metformin demonstrated significantly higher \u003cem\u003ePRKAG2\u003c/em\u003e expression across visits compared to those who were not taking metformin (t = 2.39, p = 0.018). Expression trajectories differ significantly between the metformin positive and negative groups (χ² = 12.42, p = 0.006). Top panels: \u003cem\u003ePPARG-PRKAG2\u003c/em\u003e co-expression was stable and consistently positive over time in individuals taking metformin but was heterogeneous across time in those not taking metformin, with shifts between positive and negative partial effects. Each surface was adjusted for covariates. \u0026nbsp;Bottom panels: partial effects of smoothed covariates. The solid line indicates the estimated smooth function with the shaded area representing the 95% confidence interval.\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-7907793/v1/2978479287286e8adee32ad7.png"},{"id":95528099,"identity":"60749f35-39e9-45c6-a3d6-c936a10d5b95","added_by":"auto","created_at":"2025-11-10 10:15:35","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1721987,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7907793/v1/abb093c6-1748-4c3a-b0a6-b60430244059.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"\u003cp\u003eTemporal dysregulation of\u003cem\u003e PPARG-PRKAG2\u003c/em\u003e co-expression in gray matter: Implications for cognitive decline and intervention targets in type 2 diabetes\u003c/p\u003e","fulltext":[{"header":"Background","content":"\u003cp\u003eIndividuals with type 2 diabetes (T2DM) are at significantly increased risk for neurodegeneration and associated cognitive decline, including higher incidence of Alzheimer\u0026rsquo;s disease (AD). Potential mechanisms for T2DM-related neurodegeneration include metabolic dysfunction, neuroinflammation, cerebrovascular disease, disrupted gut microbiome, and insulin resistance (\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e, \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e). T2DM and AD also share several common molecular risk factors involved in metabolism, insulin signaling, inflammation, and toxin clearance (\u003cspan additionalcitationids=\"CR4\" citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e). Many of the genes that regulate these processes are also critical for neuronal development and health.\u003c/p\u003e\u003cp\u003eUsing non-invasive imaging transcriptomics, we previously demonstrated that genes involved in insulin signaling, glucose metabolism (\u003cem\u003eIRS1, AKT1, PPARG, PRKAG2\u003c/em\u003e) and neurotransmission (\u003cem\u003eGRIN2B\u003c/em\u003e) were significantly expressed in regions of T2DM-related gray matter atrophy (\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e). Gene expression patterns interacted in both T2DM and non-diabetic control groups to predict cognitive function. However, these patterns were significantly altered in T2DM, with \u003cem\u003ePPARG\u003c/em\u003e (peroxisome proliferator-activated receptor gamma) taking over normal directional pathways from other genes to influence cognitive function. Specifically, higher \u003cem\u003ePPARG\u003c/em\u003e was associated with higher \u003cem\u003ePRKAG2\u003c/em\u003e (protein kinase AMP-activated non-catalytic subunit gamma 2) expression which in turn was associated with lower Clinical Dementia Rating (CDR) in T2DM (i.e., +PPARG \u0026loz; +PRKAG2 \u0026loz; -CDR). Given that this was a significant difference in gene network structure compared to controls, increased \u003cem\u003ePPARG\u003c/em\u003e influence may represent a compensatory mechanism in the context of T2DM neuropathology. The alteration in gene pathways occurred despite relatively preserved neuropsychological function, equivalent amyloid burden compared to controls, and lower apolipoprotein allele 4 (\u003cem\u003eAPOE4\u003c/em\u003e) risk. Thus, these molecular changes may represent an early compensatory change in T2DM (\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eHowever, our prior study was cross-sectional, and although we utilized Bayesian network analysis which determines a directional pathway model analogous to structural equation modeling, our findings were fundamentally associative. We therefore conducted a follow-up study that applied an accelerated longitudinal design in a larger sample. This approach allows for modeling temporal trajectories by combining cross-sectional data from different cohorts, each observed at a different point in time. Although each participant was observed only once, time was treated as a continuous variable, and individuals were evaluated across timepoints to approximate a longitudinal pattern. This method provides insight into group-level temporal effects, but not individual change. Thus, this represents a proof-of-concept study to support further research.\u003c/p\u003e\u003cp\u003eWe again utilized imaging transcriptomics to non-invasively evaluate gene expression patterns associated with T2DM neuropathology. Neuroimaging data is spatially co-localized with gene expression patterns from transcriptomic brain atlases to facilitate the study of in vivo molecular mechanisms underlying brain health and disease (\u003cspan additionalcitationids=\"CR8 CR9\" citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e). By linking gene expression patterns directly to brain structure and function, imaging transcriptomics opens new avenues for exploring the molecular mechanisms of neurological diseases and for developing targeted therapies. Several imaging transcriptomics studies have been conducted in neurodegenerative diseases (\u003cspan additionalcitationids=\"CR12\" citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e) but very few have involved T2DM (\u003cspan additionalcitationids=\"CR15 CR16\" citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e). Based on our prior findings, we hypothesized that co-expression of \u003cem\u003ePPARG\u003c/em\u003e and \u003cem\u003ePRKAG2\u003c/em\u003e would change over time in T2DM compared to controls. We expected that T2DM would display early compensatory activation followed by later failure consistent with prior evidence regarding neurologic compensation in neurodegeneration (\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e).\u003c/p\u003e"},{"header":"Methods","content":"\u003cp\u003e\u003cstrong\u003eParticipants\u003c/strong\u003e\u003cp\u003eWe obtained data from the Mayo Clinic Study of Aging (MCSA) via the Laboratory of Neuroimaging\u0026rsquo;s Image and Data Archive at the University of Southern California. The MCSA is a prospective, epidemiologic study of cognitive aging that began in 2004 in Olmsted County, Minnesota (\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e). The Mayo Clinic and Olmsted Medical Center institutional review boards approved the parent study from which this data was extracted and gave approval for the sharing of anonymized data. All participants provided written informed consent during the parent study. For the current study, privacy rights of participants were observed, and no personally identifiable information was obtained. The MCSA is a longitudinal study, but only cross-sectional data is currently available in the Image and Data Archive. However, participants in the cross-sectional cohort were assessed at different time points (i.e., no repeated measures; cross-sectional data at multiple timepoints from different individuals). Thus, these data lend themselves well to an accelerated longitudinal design for approximating a longitudinal trajectory across individuals. Visits were separated by approximately 1 year across 10 years. Table\u0026nbsp;1 indicates the sample sizes at each Visit. Given that visits 10 and 11 had only one participant\u0026rsquo;s data available, these visits were excluded from the analyses.\u003c/p\u003e\u003c/p\u003e\u003cp\u003eDuring the MCSA, participants underwent a medical history review, neurological evaluation, and provided demographic information. Trained nurses reviewed medical records to obtain data on chronic conditions including diagnosis of T2DM (\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e, \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e). At the time of our study, there were N\u0026thinsp;=\u0026thinsp;22,239 cases in the Image and Data Archive, but only 1,802 had magnetic resonance imaging (MRI) data and N\u0026thinsp;=\u0026thinsp;271 of these had T2DM. During the MCSA, participants were administered several neurological and neuropsychological assessments. For this study, we examined data from the CDR given our prior results (\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e). Additionally, participants underwent genotyping to determine \u003cem\u003eAPOE4\u003c/em\u003e status (positive or negative).\u003c/p\u003e\u003cp\u003e\u003cem\u003eGray Matter Volume Measurement.\u003c/em\u003e Anatomic brain MRIs were performed on all consenting individuals with no contraindications. Those who underwent MRI were not significantly different demographically from those who did not, per previous report (\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e). Imaging data were collected with a 3 Tesla GE scanner (GE Medical Systems, Milwaukee, WI) using a 3D magnetization prepared rapid acquisition gradient echo sequence (repetition time\u0026thinsp;=\u0026thinsp;2300 ms, echo time\u0026thinsp;=\u0026thinsp;3 ms, inversion time\u0026thinsp;=\u0026thinsp;900 ms, flip angle\u0026thinsp;=\u0026thinsp;8\u0026deg;, field of view\u0026thinsp;=\u0026thinsp;26 cm, in-plane matrix\u0026thinsp;=\u0026thinsp;256 \u0026times; 256, phase field of view\u0026thinsp;=\u0026thinsp;0.94, slice thickness\u0026thinsp;=\u0026thinsp;1.2 mm).\u003c/p\u003e\u003cp\u003eWe measured gray matter volumes from anatomic brain MRI using voxel-based morphometry implemented in Statistical Parametric Mapping (SPM) 12 (Wellcome Trust Centre for Neuroimaging, London, UK) (\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e). Successful segmentation and normalization were confirmed using visual and quantitative quality assurance methods. We restricted gray matter measurement to regions of atrophy that were observed in our prior study.\u003c/p\u003e\u003cp\u003e\u003cem\u003eImaging Transcriptomics\u003c/em\u003e. We obtained brain transcriptome data from the Allen Human Brain Atlas (AHBA), which is currently considered the most comprehensive transcriptional brain map available (\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e). AHBA was developed using six adult human donor brains to provide expression data from tens of thousands of genes measured from thousands of brain regions (\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e). To determine the co-localization of brain imaging and transcription data, we used the Multimodal Environment for Neuroimaging and Genomic Analysis toolbox v3.1 (\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e) to correlate gray matter volume with patterns of AHBA expression from the \u003cem\u003ePPARG\u003c/em\u003e and \u003cem\u003ePRKAG2\u003c/em\u003e. Principal Components Analysis was performed on a region-by-gene expression matrix to identify components explaining at least 95% variance across the AHBA donors. The component scores were then entered as independent variables into a multiple linear regression analysis with the corresponding gray matter volume values as the dependent variable. The accuracy of the probe to gene mapping was verified and probe data were normalized to z-scores. Representative probes were then selected in a data-driven manner considering between-donor homogeneity and the distributions of probe data. Genomic autocorrelations were calculated using an intraclass correlation coefficient (ICC) to measure the gene expression reliability across donors.\u003c/p\u003e\u003cp\u003e\u003cem\u003eStatistical Analyses.\u003c/em\u003e Demographics were compared between groups at each visit using ANOVA or Chi squared tests, as appropriate. R squared values for \u003cem\u003ePPARG\u003c/em\u003e and \u003cem\u003ePRKAG2\u003c/em\u003e were extracted for each participant. These values represent the strength of the image-transcriptome co-localization. As shown in Table\u0026nbsp;1, data density decreased substantially after visit 5, but restricting analyses to visits 1\u0026ndash;4 would have excluded later stages of the trajectory and biased inferences toward earlier phases. Thus, to evaluate the co-expression trajectory of \u003cem\u003ePPARG\u003c/em\u003e and \u003cem\u003ePRKAG2\u003c/em\u003e across study visits, we used generalized additive models (GAMs), which are well-suited for irregular or sparse longitudinal data (\u003cspan additionalcitationids=\"CR27\" citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e). GAMs stabilize estimates at timepoints with limited observations by borrowing strength from neighboring values through smooth basis functions, while still allowing for localized deviations where sufficient data exist. To prevent overfitting, particularly in sparsely sampled later visits, we used a basis dimension parameter (k\u0026thinsp;=\u0026thinsp;5), which controls the complexity of the smooth and acts as a regularization term.\u003c/p\u003e\u003cp\u003eWe evaluated whether the relationship between visit and \u003cem\u003ePPARG\u003c/em\u003e expression in predicting \u003cem\u003ePRKAG2\u003c/em\u003e differed between groups by comparing two GAMs: one with a shared smooth surface for visit across all participants (joint-smooth model) and another with separate smooths stratified by T2DM group (stratified-smooth model). Both models included fixed effects for T2DM status, a smooth term for age, and linear terms for biological sex and body mass index (BMI). Models were compared using an approximate likelihood ratio test, given that GAMs use penalized likelihood. We additionally reported the change in Akaike Information Criterion (AIC) to aid interpretation.\u003c/p\u003e\u003cp\u003eWe conducted a supplemental analysis using a traditional linear regression model to further support the GAM findings given the sparse data in later visits. We divided the cohort into visit stages with \u0026ldquo;early\u0026rdquo; stage defined as visits 1\u0026ndash;2 (1039 controls, 175 T2DM) and \u0026ldquo;late\u0026rdquo; stage (396 controls, 96 T2DM) as Visits 3 and higher. This division was based on the results from the GAM (see below for details), which indicated that in T2DM, the strongest co-expression between \u003cem\u003ePPARG\u003c/em\u003e and \u003cem\u003ePRKAG2\u003c/em\u003e occurred at Visits 1 and 2. At Visit 3, the marginal effect of \u003cem\u003ePPARG\u003c/em\u003e on \u003cem\u003ePRKAG2\u003c/em\u003e weakened. The linear regression modeled \u003cem\u003ePRKAG2\u003c/em\u003e expression from the interaction between \u003cem\u003ePPARG\u003c/em\u003e, visit stage, and T2DM status (i.e., \u003cem\u003ePPARG\u003c/em\u003e x visit stage x T2DM) with age, biological sex, and BMI as additional covariates.\u003c/p\u003e\u003cp\u003eTo examine whether the co-expression of \u003cem\u003ePPARG\u003c/em\u003e and \u003cem\u003ePRKAG2\u003c/em\u003e was associated with cognitive function and whether this relationship differed by T2DM status, we fitted a GAM with CDR sum as the dependent variable. The model included a group-stratified bivariate smooth interaction between \u003cem\u003ePPARG\u003c/em\u003e and \u003cem\u003ePRKAG2\u003c/em\u003e expression, allowing the gene-gene association to vary flexibly between T2DM and controls. Covariates included T2DM group, a smooth term for visit to capture nonlinear temporal trends, a smooth term for age, and linear terms for biological sex and BMI.\u003c/p\u003e\u003cp\u003eFinally, we compared \u003cem\u003ePRKAG2\u003c/em\u003e expression in the T2DM group between those who were or were not taking metformin. The GAM included \u003cem\u003ePPARG\u003c/em\u003e, metformin (yes/no), age, sex, and BMI.\u003c/p\u003e\u003cp\u003eData analyses were conducted in the R Statistical Package (v4.5.1, R Foundation, Vienna, Austria). GAM models were conducted using the \u0026ldquo;mgcv\u0026rdquo; library with restricted maximum likelihood for primary models and maximum likelihood for model comparisons. Alpha level for all analyses was p\u0026thinsp;\u0026lt;\u0026thinsp;0.05.\u003c/p\u003e"},{"header":"Results","content":"\u003cp\u003eMean age across groups was 65.94 +/- 8.55 years at visit 1 and 85.37 +/-2.3 years at visit 9 representing a 20-year age span across approximately 9 years of assessments (Table\u0026nbsp;1). There were no significant group-by-visit effects in terms of age (F\u0026thinsp;=\u0026thinsp;1.001, p\u0026thinsp;=\u0026thinsp;0.428), education (F\u0026thinsp;=\u0026thinsp;1.45, p\u0026thinsp;=\u0026thinsp;0.169), sex (χ\u0026sup2; = 10.46, p\u0026thinsp;=\u0026thinsp;0.234), or \u003cem\u003eAPOE4\u003c/em\u003e status (χ\u0026sup2; = 3.18, p\u0026thinsp;=\u0026thinsp;0.074). BMI was higher in the T2DM group during earlier visits (F\u0026thinsp;=\u0026thinsp;2.56, p\u0026thinsp;=\u0026thinsp;0.009).\u003c/p\u003e\u003cp\u003e\u003cem\u003ePRKAG2\u003c/em\u003e expression changed significantly over time for T2DM (F\u0026thinsp;=\u0026thinsp;69.99, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001) and controls (F\u0026thinsp;=\u0026thinsp;270.23, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001). The difference in trajectories was significant (χ\u0026sup2; = 13.82, p\u0026thinsp;=\u0026thinsp;0.001, Δ AIC = -9.93). Male sex (t\u0026thinsp;=\u0026thinsp;2.49, p\u0026thinsp;=\u0026thinsp;0.013) and smoothed age (F\u0026thinsp;=\u0026thinsp;16.71, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001) were significant contributors to the model. In T2DM, \u003cem\u003ePPARG\u003c/em\u003e strongly predicted higher \u003cem\u003ePRKAG2\u003c/em\u003e in early stages. This effect weakened or plateaued in later visits. In controls, the relationship between \u003cem\u003ePPARG\u003c/em\u003e and \u003cem\u003ePRKAG2\u003c/em\u003e was stable over time (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e).\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eThe supplemental linear regression model of early vs. late stage supported the GAM results. The \u003cem\u003ePPARG\u003c/em\u003e x T2DM interaction was significant and positive (estimate\u0026thinsp;=\u0026thinsp;0.121, p\u0026thinsp;=\u0026thinsp;0.002) indicating that \u003cem\u003ePPARG\u003c/em\u003e predicts higher \u003cem\u003ePRKAG2\u003c/em\u003e in T2DM in the early stage. The \u003cem\u003ePPARG\u003c/em\u003e x stage x T2DM interaction was significant and negative (estimate = -0.140, p\u0026thinsp;=\u0026thinsp;0.039) suggesting that the relationship between \u003cem\u003ePPARG\u003c/em\u003e and \u003cem\u003ePRKAG2\u003c/em\u003e declines (or reverses) in the late stage. The stage x T2DM interaction was significant and positive (estimate\u0026thinsp;=\u0026thinsp;0.041, p\u0026thinsp;=\u0026thinsp;0.048) indicating that individuals with T2DM in the late stage have higher baseline \u003cem\u003ePRKAG2\u003c/em\u003e, controlling for \u003cem\u003ePPARG\u003c/em\u003e.\u003c/p\u003e\u003cp\u003e\u003cem\u003ePPARG\u003c/em\u003e-\u003cem\u003ePRKAG2\u003c/em\u003e co-expression significantly influenced CDR in controls (F\u0026thinsp;=\u0026thinsp;3.17, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001) but was not significant for T2DM (F\u0026thinsp;=\u0026thinsp;7.72, p\u0026thinsp;=\u0026thinsp;0.299). Visualization of the model (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e) indicated a non-monotonic (inverted U-shaped) relationship between gene co-expression and CDR in T2DM. Alternatively, controls demonstrated a more stable, linear association with CDR. Age was a significant contributor to the model (F\u0026thinsp;=\u0026thinsp;16.70, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001) but sex and BMI were not significant (p\u0026thinsp;\u0026gt;\u0026thinsp;0.203).\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eIndividuals with T2DM who reported taking metformin demonstrated significantly higher \u003cem\u003ePRKAG2\u003c/em\u003e expression across visits compared to those who were not taking metformin (t\u0026thinsp;=\u0026thinsp;2.39, p\u0026thinsp;=\u0026thinsp;0.018). Expression changed over time in the metformin positive (F\u0026thinsp;=\u0026thinsp;39.02, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001) and metformin negative (F\u0026thinsp;=\u0026thinsp;32.45, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001) groups and these trajectories were significantly different (χ\u0026sup2; = 12.42, p\u0026thinsp;=\u0026thinsp;0.006, Δ AIC = -4.62). As shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e, gene co-expression was stable and positive for the metformin group but was less stable, fluctuating over time, and showing more variability in direction and magnitude in the metformin negative group.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eWe previously observed significant alteration in the brain transcriptome of individuals with T2DM compared to controls with a potential compensatory associative pathway from \u003cem\u003ePPARG\u003c/em\u003e to \u003cem\u003ePRKAG2\u003c/em\u003e supporting cognitive function (\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e). In the present study, we aimed to further evaluate this relationship by examining the pseudo-longitudinal trajectory of \u003cem\u003ePPARG-PRKAG2\u003c/em\u003e co-expression in gray matter tissue and the effect of this trajectory on cognitive function. In T2DM, \u003cem\u003ePPARG\u003c/em\u003e strongly predicted higher \u003cem\u003ePRKAG2\u003c/em\u003e in early stages, consistent with a compensatory mechanism. This effect weakened or plateaued in later visits, possibly due to loss of regulatory control or biological exhaustion. In controls, the relationship between \u003cem\u003ePPARG\u003c/em\u003e and \u003cem\u003ePRKAG2\u003c/em\u003e was stable over time, likely reflecting normal regulatory function without disease-driven compensation or failure.\u003c/p\u003e\u003cp\u003eThis finding was supported by supplemental stage-based modeling which indicated that, in early-stage T2DM, higher \u003cem\u003ePPARG\u003c/em\u003e was associated with higher \u003cem\u003ePRKAG2\u003c/em\u003e expression, consistent with compensatory upregulation. In late-stage T2DM, this compensatory relationship reversed, possibly reflecting system failure or overload. These findings are consistent with prior evidence regarding functional compensation in the context of neuropathology. The brain is often capable of reorganization to counteract decline and preserve function. However, ongoing pathology tends to overwhelm these compensatory mechanisms, resulting in functional decline (\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e).\u003c/p\u003e\u003cp\u003e\u003cem\u003ePPARG\u003c/em\u003e encodes peroxisome proliferator-activated receptor gamma, a transcription factor central to lipid metabolism, adipogenesis, and insulin sensitivity (\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e). \u003cem\u003ePPARG\u003c/em\u003e activation downregulates NF-κB signaling, reducing neuroinflammation, which is a hallmark of both aging and T2DM-associated cognitive decline (\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e, \u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e). Additionally, upregulation of \u003cem\u003ePPARG\u003c/em\u003e may counteract oxidative stress and insulin resistance in the brain (\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e). \u003cem\u003ePRKAG2\u003c/em\u003e is a regulatory subunit of AMP-activated protein kinase (AMPK), a master energy sensor that restores energy balance by stimulating catabolic pathways and inhibiting anabolic processes (\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e). AMPK activation is critical in maintaining ATP levels during metabolic stress, which is common in both aging brains and T2DM (\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e). \u003cem\u003ePRKAG2\u003c/em\u003e is also involved in the clearance of damaged proteins and toxins, which accumulate during neurodegeneration. Accordingly, \u003cem\u003ePRKAG2\u003c/em\u003e expression has been shown to be elevated in AD and is associated with cognitive impairment in T2DM (\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e, \u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eWith aging or disease, neurons experience reduced mitochondrial efficiency, increased oxidative stress, and altered glucose metabolism. A rise in \u003cem\u003ePPARG\u0026ndash;PRKAG2\u003c/em\u003e co-expression may represent a compensatory adaptation with \u003cem\u003ePPARG\u003c/em\u003e driving transcriptional programs that enhance lipid utilization, and AMPK signaling helping to restore energy balance. Their co-expression could synergistically buffer against metabolic decline in the aging or T2DM brain. We showed that both T2DM and control groups demonstrated increased \u003cem\u003ePRKAG2\u003c/em\u003e expression over time consistent with normal age-related changes. However, T2DM showed significantly higher \u003cem\u003ePRKAG2\u003c/em\u003e expression in later stages despite de-coupling with \u003cem\u003ePPARG\u003c/em\u003e, potentially reflecting dysregulation of compensatory processes or accelerated aging. Prior studies have demonstrated evidence of accelerated cortical brain aging in individuals with T2DM (\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e, \u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e) but our findings represent the first evidence of accelerated molecular brain aging.\u003c/p\u003e\u003cp\u003e\u003cem\u003ePPARG\u003c/em\u003e agonists, especially pioglitazone, show protective actions in experimental models of AD and diabetic encephalopathy (\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e). In rodent studies, pioglitazone reversed beta-amyloid\u0026ndash;induced memory deficits and restored learning functions (\u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e, \u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e). In human trials involving patients with mild AD or amnestic mild cognitive impairment, pioglitazone improved cognitive scores (\u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e, \u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e). \u003cem\u003ePPARG\u003c/em\u003e agonists work in part by reducing amyloid-beta levels, lowering inflammatory and oxidative markers, stabilizing brain-derived neurotrophic factor, and improving measures of insulin sensitivity (\u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e, \u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e43\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eHowever, our findings indicate that upregulation of \u003cem\u003ePPARG\u003c/em\u003e may shift from adaptive to maladaptive over time in the T2DM brain. Prior research has shown that excessive or prolonged \u003cem\u003ePPARG\u003c/em\u003e activation is linked to metabolic dysregulation such as overeating and weight gain through neuronal pathways regulating energy balance (\u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e44\u003c/span\u003e, \u003cspan citationid=\"CR45\" class=\"CitationRef\"\u003e45\u003c/span\u003e). In our cohort, participants with T2DM initially met criteria for obesity (BMI\u0026thinsp;\u0026gt;\u0026thinsp;30), but later declined below this threshold, which may reflect early \u003cem\u003ePPARG\u003c/em\u003e upregulation followed by later downregulation or improved diabetes management. Yet BMI did not contribute significantly to our gene expression or cognitive models, suggesting that \u003cem\u003ePPARG-\u003c/em\u003erelated brain effects may operate independently from adiposity. Because BMI is a coarse measure of metabolic status, factors such as insulin resistance, adipose distribution, inflammation, or diet quality may better reflect \u003cem\u003ePPARG\u003c/em\u003e-related metabolic variance. Incorporating these refined measures into future studies could clarify how peripheral metabolic remodeling and central \u003cem\u003ePPARG\u003c/em\u003e signaling interact over time to influence neurocognitive trajectories of T2DM. Further, we did not have access to blood assay data to examine the potential effects of blood glucose level or inflammatory markers on our findings. However, this research emphasizes the need for a balanced homeostatic regulation between neuroprotective and metabolic pathways to sustain cognitive function in T2DM and highlight the importance of metabolic phenotyping beyond BMI.\u003c/p\u003e\u003cp\u003eMicroRNAs regulate mRNA translation and can cross the blood brain barrier, creating the possibility that peripherally administered therapeutic interventions might influence central nervous system function. Our prior study showed that circulating miR-20b is inversely associated with longitudinal changes in fasting glucose (\u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e46\u003c/span\u003e) and \u003cem\u003ePPARG\u003c/em\u003e is a functionally validated mRNA target of miR-20b (\u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e47\u003c/span\u003e). We also showed that circulating miR-20b decreased after two years of treatment with metformin compared to an increase in expression in a placebo group, who had higher rates of incident T2DM (\u003cspan citationid=\"CR48\" class=\"CitationRef\"\u003e48\u003c/span\u003e). Because miR-20b is linked to \u003cem\u003ePPARG\u003c/em\u003e suppression, this pattern implies that higher miR-20b expression in untreated individuals may reduce \u003cem\u003ePPARG\u003c/em\u003e expression, theoretically contributing to the increased diabetes risk observed in the placebo group. These observations provide preliminary evidence that microRNAs such as miR-20b may be a therapeutic approach to modulate \u003cem\u003ePPARG-\u003c/em\u003erelated transcriptomic changes associated with neurodegeneration and cognitive decline in T2DM.\u003c/p\u003e\u003cp\u003e\u003cem\u003ePPARG\u003c/em\u003e\u0026ndash;\u003cem\u003ePRKAG2\u003c/em\u003e co-expression was significantly associated with cognitive function as measured by the CDR in controls but not in T2DM. This dissociation may imply a loss of regulatory control, breakdown of compensation, or uncoupling of biological mechanisms in T2DM. The relationship in T2DM was U-shaped where higher CDR (worse cognition) occurred at both low and high gene co-expression levels, consistent with failed compensation at extremes. These findings may also suggest that alternative neurobiologic processes are recruited to support cognitive function in T2DM. Previous studies suggest that cognitive performance is preserved in T2DM by an increase in functional brain connectivity (\u003cspan citationid=\"CR49\" class=\"CitationRef\"\u003e49\u003c/span\u003e, \u003cspan citationid=\"CR50\" class=\"CitationRef\"\u003e50\u003c/span\u003e). Functional neuroimaging was not available for this cohort in the Imaging Data Archive and therefore, future studies that include multimodal imaging transcriptomics are required.\u003c/p\u003e\u003cp\u003eMetformin treatment was associated with increased, stable \u003cem\u003ePRKAG2\u003c/em\u003e expression across visits in the T2DM group. This suggests a sustained engagement of the brain\u0026rsquo;s AMPK pathway that may contribute to metformin\u0026rsquo;s reported neuroprotective effects on neuronal metabolic equilibrium during chronic energetic stress (\u003cspan citationid=\"CR51\" class=\"CitationRef\"\u003e51\u003c/span\u003e, \u003cspan citationid=\"CR52\" class=\"CitationRef\"\u003e52\u003c/span\u003e). This stability suggests that metformin may recalibrate neuronal metabolism toward cognitive resilience, promoting continuous energy balance and proteostatic clearance through mTOR inhibition and autophagy induction (\u003cspan additionalcitationids=\"CR53\" citationid=\"CR52\" class=\"CitationRef\"\u003e52\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR54\" class=\"CitationRef\"\u003e54\u003c/span\u003e). These findings align with emerging evidence that genetic variability in PRKAG2 influences individual responses to metformin (\u003cspan citationid=\"CR55\" class=\"CitationRef\"\u003e55\u003c/span\u003e) and that metformin helps maintain blood-brain barrier integrity via AMPK activation (\u003cspan citationid=\"CR56\" class=\"CitationRef\"\u003e56\u003c/span\u003e). These observations extend the role of metformin beyond glycemic control, positioning it as a modulator of brain bioenergetic homeostasis that could delay or attenuate the molecular cascades linked to T2DM, cognitive decline, and AD (\u003cspan citationid=\"CR57\" class=\"CitationRef\"\u003e57\u003c/span\u003e). Future mechanistic and translational work should determine whether persistent \u003cem\u003ePRKAG2\u003c/em\u003e stabilization represents a biomarker of adaptive neural metabolism or a viable target for interventions aimed at preventing diabetes-related neurodegeneration.\u003c/p\u003e\u003cp\u003eThere are several limitations to this study. We utilized an accelerated longitudinal design which provides insight into group-level trajectories but not individual changes. True longitudinal studies are needed to evaluate change in brain transcriptome pathways among individuals with T2DM and the impact of these changes on cognitive function and risk for AD. We examined the co-expression of \u003cem\u003ePPARG\u003c/em\u003e and \u003cem\u003ePRKAG2\u003c/em\u003e because this was the primary altered pathway related to cognitive function in our initial study of T2DM and to reduce comparisons considering the sample size. However, alteration of other brain transcriptome pathways may also be important for cognitive function in T2DM. Data in later visits were very sparse, and we addressed this using penalized GAMs and a supplementary analysis of early versus late visits. However, analyses may have lacked power to detect alternative trajectories. While the AHBA is the most comprehensive brain transcriptome atlas to date, it was developed based on a small sample. The atlas includes gene expression patterns for thousands of brain regions but does not comprehensively cover the entire brain.\u003c/p\u003e\u003cp\u003eIn conclusion, our findings suggest that alterations of \u003cem\u003ePPARG-PRKAG2\u003c/em\u003e co-expression may be involved in the brain\u0026rsquo;s response to both normative aging and neuropathological conditions associated with T2DM. However, this response appears to become dysregulated over time in T2DM. Our results provide novel insights regarding the potential molecular mechanisms of T2DM-related cognitive decline and the risk for AD. Further research is required that includes younger adults with T2DM to determine how early these changes occur. Data regarding blood glucose levels would also provide important insights regarding the impact of T2DM disease management on the brain transcriptome. Multimodal neuroimaging will be necessary to further examine functional compensatory changes and to investigate the role of the brain transcriptome in these changes.\u003c/p\u003e"},{"header":"Abbreviations","content":"\u003cdiv class=\"DefinitionList\"\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eT2DM\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003etype 2 diabetes\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eAD\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eAlzheimer\u0026rsquo;s disease\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003ePPARG\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003ePeroxisome Proliferator-Activated Receptor Gamma\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003ePRKAG2\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eProtein Kinase AMP-Activated Non-Catalytic Subunit Gamma 2\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eMRI\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eMagnetic resonance imaging\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eAHBA\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eAllen Human Brain Atlas\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eGAM\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eGeneralized additive model\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eCDR\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eClinical Dementia Rating\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eBMI\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eBody mass index\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eAIC\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eAkaike Information Criterion\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eMCSA\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eMayo Clinic Study of Aging\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eAMPK\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eAMP-activated protein kinase\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003c/div\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003ch2\u003eEthics Approval and Consent to Participate\u003c/h2\u003e\u003cp\u003eThe Mayo Clinic and Olmsted Medical Center institutional review boards approved the parent study from which this data was extracted and gave approval for the sharing of anonymized data. All participants provided written informed consent during the parent study.\u003c/p\u003e\u003c/p\u003e\u003cp\u003e\u003cstrong\u003eConsent for Publication\u003c/strong\u003e\u003cp\u003eNot applicable.\u003c/p\u003e\u003c/p\u003e\u003cp\u003e\u003ch2\u003eCompeting Interests\u003c/h2\u003e\u003cp\u003eThe authors declare that they have no competing interests.\u003c/p\u003e\u003c/p\u003e\u003ch2\u003eFunding\u003c/h2\u003e\u003cp\u003eThe data contained in this analysis were obtained under one of the following research grants from the National Institutes of Health to the Mayo Clinic Study of Aging (U01 AG06786, Ronald Petersen, PI) or the Mayo Alzheimer\u0026rsquo;s Disease Research Center (P30 AG062677, Ronald Petersen, PI). However, the authors of this manuscript did not receive funding for this study.\u003c/p\u003e\u003ch2\u003eAuthor Contribution\u003c/h2\u003e\u003cp\u003eSRK: Conceptualization, Data curation, Formal analysis, Methodology, Project administration, Supervision, Validation, Writing - original draft, Writing - review \u0026amp; editing. KAL: Validation, Methodology, Writing - original draft, Writing - review \u0026amp; editing. HC: Methodology, Validation, Writing - original draft, Writing - review \u0026amp; editing. EF: Methodology, Validation, Writing - original draft, Writing - review \u0026amp; editing.\u003c/p\u003e\u003ch2\u003eData Availability\u003c/h2\u003e\u003cp\u003eThe dataset supporting the conclusions of this article is available to qualified researchers and industry partners from the Laboratory of Neuroimaging\u0026rsquo;s Image and Data Archive at the University of Southern California (http://loni.usc.edu) or the Global Alzheimer\u0026rsquo;s Association Interactive Network (https://www.gaain.org).\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eBlazquez E, Velazquez E, Hurtado-Carneiro V, Ruiz-Albusac JM. Insulin in the brain: its pathophysiological implications for States related with central insulin resistance, type 2 diabetes and Alzheimer's disease. Front Endocrinol (Lausanne). 2014;5:161.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLazar E, Sherzai A, Adeghate J, Sherzai D. Gut dysbiosis, insulin resistance and Alzheimer's disease: review of a novel approach to neurodegeneration. Front Biosci (Schol Ed). 2021;13(1):17\u0026ndash;29.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eBharadwaj P, Martins RN. PRKAG2 gene expression is elevated and its protein levels are associated with increased amyloid-β accumulation in the Alzheimer\u0026rsquo;s disease brain. J Alzheimer\u0026rsquo;s Disease. 2020;74(2):441\u0026ndash;8.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eRandrianarisoa E, Lehn-Stefan A, Krier J, B\u0026ouml;hm A, Heni M, De Hrabě M, et al. AMPK Subunits Harbor Largely Nonoverlapping Genetic Determinants for Body Fat Mass, Glucose Metabolism, and Cholesterol Metabolism. J Clin Endocrinol Metabolism. 2020;105(1):14\u0026ndash;25.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eShah K, DeSilva S, Abbruscato T. The role of glucose transporters in brain disease: diabetes and Alzheimer\u0026rsquo;s disease. Int J Mol Sci. 2012;13(10):12629\u0026ndash;55.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eKesler SR, Cuevas H, Lewis KA, Franco-Rocha OY, Flowers E. The expression of insulin signaling and N-methyl-D-aspartate receptor genes in areas of gray matter atrophy is associated with cognitive function in type 2 diabetes. J Alzheimers Dis. 2025:13872877251364906.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eArnatkeviciute A, Fulcher BD, Bellgrove MA, Fornito A. Imaging Transcriptomics of Brain Disorders. Biol Psychiatry Glob Open Sci. 2022;2(4):319\u0026ndash;31.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMartins D, Giacomel A, Williams SCR, Turkheimer F, Dipasquale O, Veronese M, et al. Imaging transcriptomics: Convergent cellular, transcriptomic, and molecular neuroimaging signatures in the healthy adult human brain. Cell Rep. 2021;37(13):110173.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eSunkin SM, Ng L, Lau C, Dolbeare T, Gilbert TL, Thompson CL, et al. Allen Brain Atlas: an integrated spatio-temporal portal for exploring the central nervous system. Nucleic Acids Res. 2013;41(Database issue):D996\u0026ndash;1008.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eGrothe MJ, Sepulcre J, Gonzalez-Escamilla G, Jelistratova I, Scholl M, Hansson O, et al. Molecular properties underlying regional vulnerability to Alzheimer's disease pathology. Brain. 2018;141(9):2755\u0026ndash;71.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLancour D, Dupuis J, Mayeux R, Haines JL, Pericak-Vance MA, Schellenberg GC, et al. Analysis of brain region-specific co-expression networks reveals clustering of established and novel genes associated with Alzheimer disease. Alzheimers Res Ther. 2020;12(1):1\u0026ndash;11.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eBaik JY, Kim M, Bao J, Long Q, Shen L, Initiative ADN. Identifying Alzheimer\u0026rsquo;s genes via brain transcriptome mapping. BMC Med Genom. 2022;15(Suppl 2):116.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMroczek M, Desouky A, Sirry W. Imaging transcriptomics in neurodegenerative diseases. J Neuroimaging. 2021;31(2):244\u0026ndash;50.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLiu Q, Du X, Zhang Y, Ding H, Qin W, Zhang Q. Associations between morphometric similarity network and brain gene expression in type 2 diabetes mellitus. Neuroscience. 2025;572:84\u0026ndash;92.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eZhang Y, Du X, Qin W, Fu Y, Wang Z, Zhang Q. Association between gene expression and altered resting-state functional networks in type 2 diabetes. Front Aging Neurosci. 2023;15:1290231.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMullins RJ, Diehl TC, Chia CW, Kapogiannis D. Insulin Resistance as a Link between Amyloid-Beta and Tau Pathologies in Alzheimer's Disease. Front Aging Neurosci. 2017;9:118.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHuang X, He YX, Wan S. Genetic mechanisms of hemispheric functional connectivity in diabetic retinopathy: a joint neuroimaging and transcriptomic study. Front Cell Dev Biol. 2025;13:1590627.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eScheller E, Minkova L, Leitner M, Kloppel S. Attempted and successful compensation in preclinical and early manifest neurodegeneration - a review of task FMRI studies. Front Psychiatry. 2014;5:132.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eRoberts RO, Geda YE, Knopman DS, Cha RH, Pankratz VS, Boeve BF, et al. The Mayo Clinic Study of Aging: design and sampling, participation, baseline measures and sample characteristics. Neuroepidemiology. 2008;30(1):58\u0026ndash;69.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eWennberg AMV, Hagen CE, Edwards K, Roberts RO, Machulda MM, Knopman DS, et al. Association of antidiabetic medication use, cognitive decline, and risk of cognitive impairment in older people with type 2 diabetes: Results from the population-based Mayo Clinic Study of Aging. Int J Geriatr Psychiatry. 2018;33(8):1114\u0026ndash;20.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eAlhurani RE, Vassilaki M, Aakre JA, Mielke MM, Kremers WK, Machulda MM, et al. Decline in Weight and Incident Mild Cognitive Impairment: Mayo Clinic Study of Aging. JAMA Neurol. 2016;73(4):439\u0026ndash;46.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eWhitwell JL, Wiste HJ, Weigand SD, Rocca WA, Knopman DS, Roberts RO, et al. Comparison of imaging biomarkers in the Alzheimer Disease Neuroimaging Initiative and the Mayo Clinic Study of Aging. Arch Neurol. 2012;69(5):614\u0026ndash;22.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eKurth F, Luders E, Gaser C. Voxel-Based Morphometry. Brain Mapping2015. pp. 345\u0026ndash;9.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHawrylycz MJ, Lein ES, Guillozet-Bongaarts AL, Shen EH, Ng L, Miller JA, et al. An anatomically comprehensive atlas of the adult human brain transcriptome. Nature. 2012;489(7416):391\u0026ndash;9.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eRizzo G, Veronese M, Expert P, Turkheimer FE, Bertoldo A. MENGA: A New Comprehensive Tool for the Integration of Neuroimaging Data and the Allen Human Brain Transcriptome Atlas. PLoS ONE. 2016;11(2):e0148744.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMundo AI, Tipton JR, Muldoon TJ. Generalized additive models to analyze nonlinear trends in biomedical longitudinal data using R: Beyond repeated measures ANOVA and linear mixed models. Stat Med. 2022;41(21):4266\u0026ndash;83.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHastie T, Tibshirani R. Generalized additive models. Stat Sci. 1986;1(3):297\u0026ndash;310.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eYilema SA, Shiferaw YA, Belay AT, Belay DB. Mapping the spatial disparities of HIV prevalence in Ethiopian zones using the generalized additive model. Sci Rep. 2024;14(1):6215.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eStump M, Guo D-F, Lu K-T, Mukohda M, Cassell MD, Norris AW, et al. Nervous System Expression of PPARγ and Mutant PPARγ Has Profound Effects on Metabolic Regulation and Brain Development. Endocrinology. 2016;157(11):4266\u0026ndash;75.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eZhang Y, Chen C, Jiang Y, Wang S, Wu X, Wang K. PPARγ coactivator-1α (PGC-1α) protects neuroblastoma cells against amyloid-beta (Aβ) induced cell death and neuroinflammation via NF-κB pathway. BMC Neurosci. 2017;18(1):69.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eRom S, Zuluaga-Ramirez V, Gajghate S, Seliga A, Winfield M, Heldt NA, et al. Hyperglycemia-driven neuroinflammation compromises BBB leading to memory loss in both diabetes mellitus (DM) type 1 and type 2 mouse models. Mol Neurobiol. 2019;56(3):1883\u0026ndash;96.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eWang P, Li B, Cai G, Huang M, Jiang L, Pu J et al. Activation of PPAR-γ by Pioglitazone Attenuates Oxidative Stress in Aging Rat Cerebral Arteries Through Upregulating UCP2. J Cardiovasc Pharmacol. 2014;64(6).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eBonomini F, Rodella LF, Rezzani R. Metabolic syndrome, aging and involvement of oxidative stress. Aging disease. 2015;6(2):109.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eKim E, Lee SH, Lee KS, Cheong H-K, Namkoong K, Hong CH, et al. AMPK γ2 subunit gene PRKAG2 polymorphism associated with cognitive impairment as well as diabetes in old age. Psychoneuroendocrinology. 2012;37(3):358\u0026ndash;65.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eAntal B, McMahon LP, Sultan SF, Lithen A, Wexler DJ, Dickerson B, et al. Type 2 diabetes mellitus accelerates brain aging and cognitive decline: Complementary findings from UK Biobank and meta-analyses. eLife. 2022;11:e73138.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eDove A, Wang J, Huang H, Dunk MM, Sakakibara S, Guitart-Masip M, et al. Diabetes, Prediabetes, and Brain Aging: The Role of Healthy Lifestyle. Diabetes Care. 2024;47(10):1794\u0026ndash;802.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eAlhowail A, Alsikhan R, Alsaud M, Aldubayan M, Rabbani SI. Protective Effects of Pioglitazone on Cognitive Impairment and the Underlying Mechanisms: A Review of Literature. Drug Des Devel Ther. 2022;16:2919\u0026ndash;31.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eSharma B, Singh N. Behavioral and biochemical investigations to explore pharmacological potential of PPAR-gamma agonists in vascular dementia of diabetic rats. Pharmacol Biochem Behav. 2011;100(2):320\u0026ndash;9.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLiu J, Wang LN, Jia JP. Peroxisome proliferator-activated receptor-gamma agonists for Alzheimer's disease and amnestic mild cognitive impairment: a systematic review and meta-analysis. Drugs Aging. 2015;32(1):57\u0026ndash;65.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eWatson GS, Cholerton BA, Reger MA, Baker LD, Plymate SR, Asthana S, et al. Preserved cognition in patients with early Alzheimer disease and amnestic mild cognitive impairment during treatment with rosiglitazone: a preliminary study. Am J Geriatr Psychiatry. 2005;13(11):950\u0026ndash;8.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMu\u0026ntilde;oz-Jim\u0026eacute;nez M, Zaarkti A, Garc\u0026iacute;a-Arn\u0026eacute;s JA, Garc\u0026iacute;a-Casares N. Antidiabetic Drugs in Alzheimer's Disease and Mild Cognitive Impairment: A Systematic Review. Dement Geriatr Cogn Disord. 2020;49(5):423\u0026ndash;34.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003ePrakash A, Kumar A. Role of nuclear receptor on regulation of BDNF and neuroinflammation in hippocampus of β-amyloid animal model of Alzheimer's disease. Neurotox Res. 2014;25(4):335\u0026ndash;47.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eTitus C, Hoque MT, Bendayan R. PPAR agonists for the treatment of neuroinflammatory diseases. Trends Pharmacol Sci. 2024;45(1):9\u0026ndash;23.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eDiano S, Liu Z-W, Jeong JK, Dietrich MO, Ruan H-B, Kim E, et al. Peroxisome proliferation\u0026ndash;associated control of reactive oxygen species sets melanocortin tone and feeding in diet-induced obesity. Nat Med. 2011;17(9):1121\u0026ndash;7.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eStump M, Guo DF, Lu KT, Mukohda M, Liu X, Rahmouni K, et al. Effect of selective expression of dominant-negative PPARγ in pro-opiomelanocortin neurons on the control of energy balance. Physiol Genomics. 2016;48(7):491\u0026ndash;501.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eFlowers E, Allen IE, Kanaya AM, Aouizerat BE. Circulating MicroRNAs predict glycemic improvement and response to a behavioral intervention. Biomark Res. 2021;9(1):65.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eCui S, Yu S, Huang HY, Lin YC, Huang Y, Zhang B, et al. miRTarBase 2025: updates to the collection of experimentally validated microRNA-target interactions. Nucleic Acids Res. 2025;53(D1):D147\u0026ndash;56.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLewis KA, Stroebel BM, Zhang L, Aouizerat B, Mattis AN, Flowers E. MicroRNAs Associated with Metformin Treatment in the Diabetes Prevention Program. Int J Mol Sci. 2024;25(11).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eWang J, Zhou S, Deng D, Chen M, Cai H, Zhang C, et al. Compensatory thalamocortical functional hyperconnectivity in type 2 Diabetes Mellitus. Brain Imaging Behav. 2022;16(6):2556\u0026ndash;68.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eZhou B, Wang X, Yang Q, Wu F, Tang L, Wang J et al. Topological Alterations of the Brain Functional Network in Type 2 Diabetes Mellitus Patients With and Without Mild Cognitive Impairment. Front Aging Neurosci. 2022;Volume 14\u0026ndash;2022:834319.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eKim HI, Han Y, Park J. AMP-Activated Protein Kinases in Health and Disease. Int J Mol Sci. 2025;26(16):8075.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHardie DG. Keeping the home fires burning: AMP-activated protein kinase. J R Soc Interface. 2018;15(138).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHinson JT, Chopra A, Lowe A, Sheng CC, Gupta RM, Kuppusamy R, et al. Integrative Analysis of PRKAG2 Cardiomyopathy iPS and Microtissue Models Identifies AMPK as a Regulator of Metabolism, Survival, and Fibrosis. Cell Rep. 2016;17(12):3292\u0026ndash;304.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eWu S, Zou M-HAMPK. Mitochondrial Function, and Cardiovascular Disease. Int J Mol Sci [Internet]. 2020; 21(14).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eXiao D, Liu J-Y, Zhang S-M, Liu R-R, Yin J-Y, Han X-Y et al. A two-stage study identifies two novel polymorphisms in PRKAG2 affecting metformin response in Chinese type 2 diabetes patients. Pharmacogenomics Personalized Med. 2021:745\u0026ndash;55.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLi N, Zhou T, Fei E. Actions of Metformin in the Brain: A New Perspective of Metformin Treatments in Related Neurological Disorders. Int J Mol Sci [Internet]. 2022; 23(15).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMu\u0026ntilde;oz-Jim\u0026eacute;nez M, Zaarkti A, Garc\u0026iacute;a-Arn\u0026eacute;s JA, Garc\u0026iacute;a-Casares N. Antidiabetic Drugs in Alzheimer\u0026rsquo;s Disease and Mild Cognitive Impairment: A Systematic Review. Dement Geriatr Cogn Disord. 2020;49(5):423\u0026ndash;34.\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"},{"header":"Tables","content":"\u003cp\u003eTable 1.\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eVisit\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eGroup\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eN\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eAge\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eMale\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eBMI\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eCDR\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eEducation\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eAPOE4\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eMetformin\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003ePPARG\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003ePRKAG2\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eControl\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e724\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e65.47 (8.4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e377 (52%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e28.37 (4.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.103 (0.49)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e14.88 (2.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e216 (30%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e2 (0.28%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.325 (0.04)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.294 (0.03)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eT2DM\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e121\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e68.70 (8.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e70 (58%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e31.62 (5.2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.218 (0.74)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e13.92 (2.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e27 (22%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e73 (60%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.312 (0.05)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.288 (0.04)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eControl\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e315\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e68.44 (9.2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e152 (48%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e28.48 (5.0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.102 (0.41)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e15.03 (2.6)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e97 (31%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e1(0.32%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.319 (0.04)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.289 (0.03)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eT2DM\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e54\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e71.51 (8.0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e33 (61%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e30.26 (3.7)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.269 (0.70)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e14.61 (2.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e16 (30%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e32(59%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.312 (0.03)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.286 (0.03)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eControl\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e113\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e73.72 (8.6)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e57 (50%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e27.22 (4.6)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.296 (1.14)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e14.92 (2.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e34 (30%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0(0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.313 (0.04)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.283 (0.03)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eT2DM\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e33\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e73.64 (8.0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e20 (61%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e33.74 (5.6)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.273 (0.55)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e13.91 (2.7)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e8 (24%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e21(64%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.298 (0.04)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.280 (0.03)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eControl\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e155\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e78.98 (5.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e78 (50%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e27.30 (4.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.386 (1.10)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e14.24 (3.0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e43 (28%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0(0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.306 (0.04)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.279 (0.03)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eT2DM\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e38\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e79.32 (4.8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e28 (74%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e29.10 (5.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.395 (1.16)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e14.26 (2.6)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e6 (16%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e15(40%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.307 (0.05)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.275 (0.03)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eControl\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e83\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e81.99 (4.0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e48 (58%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e26.82 (4.4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.470 (1.28)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e14.60 (3.1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e19 (23%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0(0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.299 (0.04)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.275 (0.03)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eT2DM\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e14\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e81.89 (3.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e9 (64%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e29.89 (6.8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.107 (0.21)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e14.57 (2.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e5 (36%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e8(57%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.293 (0.03)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.269 (0.03)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eControl\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e18\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e81.91 (3.7)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e8 (44%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e27.51 (4.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.806 (1.61)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e13.78 (2.6)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e4 (22%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0(0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.284 (0.04)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.273 (0.03)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eT2DM\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e84.16 (3.6)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e3 (60%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e29.26 (4.4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.900 (1.25)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e12.20 (3.0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e3 (60%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e1(20%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.299 (0.03)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.270 (0.01)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e7\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eControl\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e8\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e83.31 (3.6)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e3 (37.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e27.92 (4.7)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.188 (0.53)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e14.25 (2.6)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e4 (50%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0(0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.288 (0.03)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.277 (0.03)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eT2DM\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e78.80 (N/A)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e1 (100%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e25.77 (N/A)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eN/A (N/A)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e16.00 (N/A)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e1 (100%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e1(100%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.237 (N/A)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.250 (N/A)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e8\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eControl\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e14\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e84.74 (2.2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e10 (71%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e26.73 (3.4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.214 (0.38)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e15.00 (3.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e4 (29%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0(0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.314 (0.05)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.286 (0.03)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eT2DM\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e85.20 (3.4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e1 (33%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e26.80 (2.1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.167 (0.29)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e10.67 (2.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e3 (100%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e2(67%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.281 (0.03)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.282 (0.02)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e9\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eControl\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e85.88 (2.0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e5 (100%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e26.11 (2.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.100 (0.22)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e16.00 (3.2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e1 (20%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0(0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.308 (0.02)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.287 (0.02)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eT2DM\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e84.10 (3.4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e2 (100%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e29.32 (3.1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.000 (0.00)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e17.00 (4.2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0 (0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e1(50%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.273 (0.08)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.280 (0.04)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"type 2 diabetes, cognitive decline, imaging transcriptomics, MRI, clinical dementia rating, miRNA-20b, metformin","lastPublishedDoi":"10.21203/rs.3.rs-7907793/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7907793/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cstrong\u003eBackground: \u003c/strong\u003eType 2 diabetes mellitus (T2DM) is associated with increased risk for cognitive decline and diagnosis of Alzheimer’s disease (AD). The mechanisms of T2DM related dementia remain unclear.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMethods: \u003c/strong\u003eBrain magnetic resonance imaging was retrospectively obtained for 1,802 adults (age 66 +/- 9 years, 47% male) of whom N = 271 had T2DM. We applied an accelerated longitudinal design and imaging transcriptomics to non-invasively examine the group-level trajectories of \u003cem\u003ePPARG\u003c/em\u003eand \u003cem\u003ePRKAG2\u003c/em\u003e co-expression in gray matter.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eResults: \u003c/strong\u003eGene expression trajectories differed significantly between T2DM and controls (χ² = 13.82, p = 0.001). Co-expression was higher in early stages and then weakened in later stages among T2DM, while remaining stable over time in controls. \u003cem\u003ePPARG\u003c/em\u003e and \u003cem\u003ePRKAG2\u003c/em\u003e co-expression was significantly associated with cognitive function in controls (F = 3.17, p \u0026lt; 0.001) but not T2DM (F = 7.72, p = 0.299) suggesting dysregulated or failed compensatory mechanisms. Individuals with T2DM not taking metformin demonstrated unstable gene co-expression over time compared to those taking metformin (χ² = 12.42, p = 0.006).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConclusions: \u003c/strong\u003eThe convergence of \u003cem\u003ePPARG\u003c/em\u003e-mediated metabolic remodeling and \u003cem\u003ePRKAG2\u003c/em\u003e/AMPK-driven energy sensing may act as a coordinated neuroprotective mechanism, upregulated in response to cellular stress in both pathological (T2DM) and normative (aging) contexts. However, these processes appear to become dysregulated in T2DM, potentially resulting in cognitive decline and increased risk for dementia.\u003c/p\u003e","manuscriptTitle":"Temporal dysregulation of PPARG-PRKAG2 co-expression in gray matter: Implications for cognitive decline and intervention targets in type 2 diabetes","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-10-22 19:02:06","doi":"10.21203/rs.3.rs-7907793/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"13f2d750-c9aa-4bfb-9e5f-be4f2747012a","owner":[],"postedDate":"October 22nd, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2025-11-10T01:53:36+00:00","versionOfRecord":[],"versionCreatedAt":"2025-10-22 19:02:06","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-7907793","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7907793","identity":"rs-7907793","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}