DyNAMiC: A prospective longitudinal study of dopamine and brain connectomes: A new window into cognitive aging

Abstract Concomitant exploration of structural, functional, and neurochemical brain mechanisms underlying age‐related cognitive decline is crucial in promoting healthy aging. Here, we present the DopamiNe, Age, connectoMe, and Cognition (DyNAMiC) project, a multimodal, prospective 5‐year longitudinal study spanning the adult human lifespan. DyNAMiC examines age‐related changes in the brain’s structural and functional connectome in relation to changes in dopamine D1 receptor availability (D1DR), and their associations to cognitive decline. Critically, due to the complete lack of longitudinal D1DR data, the true trajectory of one of the most age‐sensitive dopamine systems remains unknown. The first DyNAMiC wave included 180 healthy participants (20–80 years). Brain imaging included magnetic resonance imaging assessing brain structure (white matter, gray matter, iron), perfusion, and function (during rest and task), and positron emission tomography (PET) with the [11C]SCH23390 radioligand. A subsample (n = 20, >65 years) was additionally scanned with [11C]raclopride PET measuring D2DR. Age‐related variation was evident for multiple modalities, such as D1DR; D2DR, and performance across the domains of episodic memory, working memory, and perceptual speed. Initial analyses demonstrated an inverted u‐shaped association between D1DR and resting‐state functional connectivity across cortical network nodes, such that regions with intermediate D1DR levels showed the highest levels of nodal strength. Evident within each age group, this is the first observation of such an association across the adult lifespan, suggesting that emergent functional architecture depends on underlying D1DR systems. Taken together, DyNAMiC is the largest D1DR study worldwide, and will enable a comprehensive examination of brain mechanisms underlying age‐related cognitive decline.


| INTRODUC TI ON
The world's elderly population is rapidly increasing and the number of individuals with age-related cognitive impairments is expected to double over the next 50 years (United Nations, World Population Aging Report, 2015;2019). Current knowledge of the brain mechanisms underlying age-related cognitive decline is, however, insufficient to inform effective interventions promoting healthy aging. This is largely due to a scarcity of longitudinal multimodal data, hindering a comprehensive understanding of age-related brain changes and their role in cognitive decline. In this paper, we describe the DopamiNe, Age, connectoMe, and Cognition (DyNAMiC) study, designed to examine changes in dopamine (DA) functions and in the brain's structural and functional connectome across the adult lifespan, and the extent to which such changes impact cognitive decline in aging.
The human connectome describes elements and connections forming the human brain (Kelly et al., 2012;Petersen & Sporns, 2015;Sporns, 2013Sporns, , 2014Sporns et al., 2005). The structural connectome consists of white matter (WM) tracts interconnecting brain regions and represents the brain's structural architecture, whereas the functional connectome, defined as synchronized activity across distal parts of the brain (i.e., functional connectivity; FC), reflects the functional architecture of complex neural systems. Spontaneous activity within the functional connectome at rest consumes the majority of the brain's energy (Shulman et al., 2004), acts as a fingerprint (Finn et al., 2015), shows good test-retest reliability (Zuo & Xing, 2014), and persists during sleep and anesthesia , suggesting that it is a rather stable trait of the brain. Numerous studies indicate correspondence between structural and functional connectomes (Damoiseaux & Greicius, 2009;Greicius et al., 2009;Hermundstad et al., 2013;Honey et al., 2009;Osmanlıoğlu et al., 2019;van den Heuvel et al., 2009), with structure-function associations varying across cortical regions (Baum et al., 2020;Vázquez-Rodríguez et al., 2019). On this view, age-related alterations in the functional connectome partly reflect changes in the structural connectome known to occur in aging (Betzel et al., 2014;Damoiseaux, 2017;Madden et al., 2009Madden et al., , 2020Salami et al., 2012). Granted that cross-sectional findings suggest variation in the relationship between structural and functional connectomes across the lifespan (Betzel et al., 2014), longitudinal observations are few and inconclusive (Fjell et al., 2016;Pedersen et al., 2021).
Regarding the functional connectome in aging, a recurrent cross-sectional observation is decreased within-network FC and increased between-network FC in older adults (for reviews see Damoiseaux, 2017;Zuo et al., 2017), possibly reflecting neural dedifferentiation (i.e., reduced functional network specialization) in old age (Chan et al., 2014;Geerligs et al., 2014). Although changes in within-network FC of large-scale networks, such as the default mode network (DMN; Buckner et al., 2008;Raichle et al., 2001), have been reported in longitudinal studies , evidence for increased between-network FC and changes in the overall functional architecture of the brain are limited (Ng et al., 2016;Pedersen et al., 2021). Different directions of age effects on within-and between-network FC have further been observed comparing cross-sectional and longitudinal estimates (Fjell et al., 2015).
Relatedly, cross-sectional studies report the functional connectome as predictive of cognition (Andrews-Hanna et al., 2007;Damoiseaux et al., 2008;Wang et al., 2010), whereas longitudinal evidence of a link between age-related cognitive decline and disruption in the functional connectome is limited (but see Malagurski et al., 2020;Pedersen et al., 2021).
It is possible that the integrity of the functional connectome and its link to cognition in aging is dependent on the integrity of the dopaminergic (DA) system, given that DA neurotransmission plays a key role in cognition through its modulation of synaptic activity enhancing specificity in the neuronal signal (El-Ghundi et al., 2007;Seamans & Yang, 2004). Based on meta-analyses, the two main postsynaptic DA receptor families, the DA D1 (D1DR) and D2 (D2DR) receptor families, show linear deterioration from early to late adulthood (Karrer et al., 2017; but see Seaman et al., 2019).
Critically, cross-sectional estimates of age-related alterations of D1DR are most often limited to extreme age groups, with no study covering the adult lifespan (but see Rinne et al., 1990), and a balanced number of individuals per decade. Furthermore, no longitudinal data currently exist for D1DR, reported as the most age-sensitive dopaminergic marker (Karrer et al., 2017). Thus, it remains unclear whether cross-sectional estimates represent true rate and shape of D1DR decline across the lifespan.
In young adults, D1DR is associated with blood oxygenation level-dependent (BOLD) brain activation (Turner et al., 2020) and to aspects of the functional connectome Roffman et al., 2016). Yet, to what extent the spatial configuration of the brain's functional architecture (e.g., variation in nodal FC across regions) depends on underlying DA receptor distributions (Shine Significance Simultaneous assessment of structural, functional, and neurochemical brain mechanisms underlying age-related cognitive decline is crucial in promoting healthy aging.
Longitudinal multimodal data are, however, currently lacking. We present the DyNAMiC project, which will ultimately examine changes in cognition, dopamine (DA), and the brain's structural and functional connectome across the adult lifespan. DyNAMiC constitutes the largest DA D1 receptor study worldwide and also includes DA D2 assessment for a subsample of participants. Data from the first wave of DyNAMiC provide unique opportunities to examine contributions of different brain measures to individual differences in cognition, and to identify associations between functional and molecular brain systems as potential mechanisms of cognitive decline. et al., 2019), and in what manner the relation between D1DR and the functional connectome changes across the lifespan remains unknown. In addition to addressing these questions, an important contribution of DyNAMiC will be to provide longitudinal estimates of D1DR (and the structural and functional connectome), enabling assessments of unique and shared mechanisms contributing to cognitive decline in aging. Concentrations of D1DR and D2DR vary across the brain , and DA is distributed through various distinct pathways (Haber, 2014). Accordingly, studies suggest that D1DR and D2DR may differentially contribute to prefrontalbased processes of working memory and limbic-based processes of episodic memory (Liggins, 2009;Nyberg et al., 2016;Takahashi et al., 2007Takahashi et al., , 2008. Thus, it is possible that D1DR and D2DR differentially contribute to specific aspects of the functional connectome, shaping their roles in age-related cognitive decline. Finally, brain and cognitive functions are characterized by a high degree of inter-individual heterogeneity , likely reflecting both genetic and environmental factors. Various genetic and lifestyle factors (e.g., intellectual and physical activities), as well as vascular risk factors, all contribute to individual differences in cognitive aging (Dahle et al., 2009;Köhncke et al., 2018;Mintzer et al., 2019;Papenberg et al., 2015). For instance, physical activity measured over a decade may modify decline in some brain-driven measures, including parts of the functional connectome (Boraxbekk et al., 2016).
DyNAMiC is a prospective 5-year longitudinal study in collaboration between the Umeå Center for Functional Brain Imaging (UFBI) at Umeå University and the Aging Research Center (ARC) at Karolinska Institutet/Stockholm University, Sweden. The principal aims of DyNAMiC are to (1) determine the rates and trajectories of age-related changes in different brain measures across the adult lifespan, focusing on the D1DR system and the brain connectome and; (2) delineate shared and unique contributions of changes in these brain measures to changes in various cognitive domains; and (3) identify factors contributing to altered brain integrity, such as genetic polymorphisms, health, and lifestyle factors (e.g., blood pressure, diet, and physical activity).
The specific aims of the present paper were, first, to provide a comprehensive overview of DyNAMiC and the baseline wave of data collection which included: (a) magnetic resonance imaging (MRI) for anatomical and functional brain measures; (b) PET scanning to assess postsynaptic DA systems with the D1DR radioligand

| Recruitment procedure
Participants (n = 180) were recruited from six decades from the age 20 to 80 years across the adult lifespan. Recruitment was ongoing throughout the first wave of data collection, between 2017 and 2020, with individuals included at baseline born between 1937 and 2000. Efforts were made to include approximately the same number of individuals from each decade of interest, and to achieve even distributions of age and sex (Table S1). Invitation letters were sent out to a sample randomly drawn (within each decade) from the population registry of Umeå, Sweden. The expected number of returnees for time point 2 was 120, based on previous longitudinal neuroimaging studies conducted in Umeå (COBRA: Nevalainen et al., 2015;andBetula: Nilsson et al., 1997, 2004), with attrition rates of ~30% between baseline and time point 2. Further information about the recruitment procedure is provided in the Supporting Information.
A set of exclusion criteria was implemented during recruitment to create a sample of healthy participants without conditions and medical treatment potentially affecting brain functioning and cognition. Respondents were excluded if they met one or more of the following criteria: brain injury or neurological disorder, dementia, neurodevelopmental disorder, psychiatric diagnosis, psychopharmacological treatment, history of severe head trauma, substance abuse or dependence, and illicit drug use. Individuals with other chronic or serious medical conditions (e.g., cancer, diabetes, and Parkinson's disease) were also excluded.
Additionally, respondents had to meet the prerequisites for the study procedure in order to be included. This involved being able to undergo a 90-min MRI scan, being able to see and hear adequately inside and outside the scanner environment, and being a native Swedish speaker. Thus, individuals having any non-MRI safe metal implant or residue (e.g., pacemaker, medicine pump, neural stimulator, arterial clips, prostheses, splinters, and welding sparks) or MRI-safe metal implant that might diminish image quality (e.g., titanium screw or permanent braces in the upper jaw) were excluded. Radiation safety was also taken into account before inclusion. Individuals having previously participated in a research project involving a PET scan, or who had recently undergone any other procedure involving the injection of a radioactively marked substance, were excluded. Pregnant women were also excluded, and breastfeeding women had to follow strict instructions (including no breastfeeding for at least 6 hr after the PET scan), in order to participate.
Demographic information is presented in Table 1. Three participants dropped out during data collection. Thus, DyNAMiC includes 177 participants with baseline data from both the MRI and PET sessions. All participants underwent the Mini Mental State Examination (MMSE, Folstein et al., 1975), scoring ≥26.

| Study design
DyNAMiC includes two time points of data collection. Data collection for the first time point was carried out in 2017-2020 at Umeå University Hospital. This will remain the same for the second time point, planned to follow 5 years from baseline, starting in 2022 (see Figure 1). Participants will be scheduled for testing in an order corresponding to their participation at baseline. All included procedures and testing will be the same for time point 2 as for the baseline measurement. At each wave, testing is distributed over 2 or 3 separate days, including one MRI session and one PET session (SCH-PET to assess D1DR) for the full sample, as well as a second PET session (RAC-PET for assessment of D2DR) for a subsample of participants (n = 20; >65 years of age). Testing procedures for the different sessions are outlined in Figure 1.
The first session included the MMSE, testing of specific cognitive functions, and MRI assessment, and lasted approximately 3 hr 45 min. Participants provided written informed consent at the beginning of the session. Before MRI scanning, which lasted for 90 min, they responded to a short status questionnaire assessing their current alertness with questions of sleep and caffeine intake, and practiced the in-scanner working memory n-back task.
Participants also received a comprehensive lifestyle questionnaire to fill out at home. The second session included cognitive testing, blood pressure measurement, blood sampling, and a 60 min SCH-PET scan following the individual fitting of a thermoplastic mask for in-scanner head stabilization. This session lasted approximately 2 hr 30 min in total. For the subsample of older participants, a third session included a 60 min RAC-PET scan using the fitted mask from their first PET session.
For the larger group of participants, completing only the SCH-PET session, data collection proceeded as planned for 83% (n = 130), with an average of 2 days between MRI and PET sessions (ranging from 1 to 10 days). For the remaining 17%, data collection was delayed, resulting in an average of 55 days between sessions (ranging between 18 and 141 days). For the majority of the smaller, double-PET subsample (80%, n = 16), data collection was completed with an average of 10 days in between the first and last session (ranging from 7 to 11 days). PET sessions were delayed for the remaining four participants, with the number of days between the first and last session ranging from 35 to 49. The most common reason for delay between sessions was PET tracer production failure.

| Cognitive measures
A battery of cognitive tests assessed episodic memory, working memory, and perceptual speed. This battery was originally developed for the COGITO study (Schmiedek, Bauer, et al., 2010;, and later adapted for the Umeå-based COBRA study including Swedish participants (Nevalainen et al., 2015). Each of the three cognitive domains was assessed using three separate tasks containing letter-, number-, and figure-based material, respectively ( Figure 2). Tasks were presented on a computer, in the same order across participants, which provided their responses by either typing in words or numbers; using the computer mouse; or pressing keys marked by different colors corresponding to specific response alternatives. Every task started with a written instruction, after which one or several practice runs were completed (varying across tasks). Testing then followed in several runs, resulting in the overall performance (e.g., accuracy, response times, or frequencies) being a composite of performance across the separate test runs. Testing was conducted across two sessions prior to MRI and PET scanning ( Figure 1). A test leader was present throughout both sessions. This section provides a description of the episodic memory, working memory, and perceptual speed tests, as well as included tests of semantic knowledge, implicit learning, and motor speed.
Reliability measures of cognitive tests including two trials were estimated using the Spearman-Brown coefficient, which might be less biased than Cronbach's alpha (Eisinga et al., 2013). Nevertheless, differences between Spearman-Brown coefficients and Cronbach's alpha for our measures were small (<0.01). Cronbach's alpha was used as the measure of reliability for cognitive tests including more than two trials.   participants were presented with 16 words that appeared one by one on the computer screen. Words were concrete Swedish nouns (e.g., flower) and no two words shared the same first three letters.

| Episodic memory
During the first phase, participants encoded each word for 6 s, with an inter-stimulus interval (ISI) of 1 s. Following the presentation of all words in the series, participants used the keyboard to type in as many of the presented words that could recall, in any order. Performance was defined as the number of correctly recalled words. Two trials of this test were completed after an initial practice trial, yielding the maximum score of 32. The reliability of this measure across the two trials was 0.86 (Spearman-Brown coefficient, based on n = 180 subjects).
In the number-word recall test, participants were required to memorize pairs of two-digit numbers and concrete plural nouns (e.g., 46 dogs). Ten number-word pairs were presented consecutively, each displayed for 6 s, with an ISI of 1 s. Retrieval immediately followed, in which every word was consecutively presented again, but in a different order than during encoding. For each word, participants had to recall the associated two-digit number, and type the answer using the keyboard. A response was required for each word, meaning that participants had to provide a guessing-based response even if they did not recall the correct number. Following an initial practice trial, this test was administered in two trials with a total maximum score of 20 correctly recalled numbers. This test showed a reliability of 0.76 (Spearman-Brown coefficient, n = 180) across trials.
In the object-location memory task, participants encoded objects presented on different locations in a 6 × 6 square grid displayed on the computer screen. Each encoding trial involved 12 objects, one by one, in distinct locations within the grid. Each object-position pairing was displayed for 8 s before disappearing, with an ISI of 1 s. Directly following encoding, all objects were simultaneously displayed next to the grid for participants to move them (in any order) to their correct location in the grid. If unable to recall an object's correct position, participants had to guess and place the object at a location to the best of their ability. Two test trials of this task were administered after a practice trial, yielding a total maximum score of 24. The reliability of this measure was 0.69 (Spearman-Brown coefficient, n = 180).

| Working memory
Working memory was also tested using three tasks, letter updating, number updating, and spatial updating ( Figure 2). These three tests were different from the working memory n-back task performed by participants during fMRI scanning (described in Section 2.8.1.5.).
During letter updating, participants were presented with a sequence of capital letters (A-D), consecutively on the computer screen, requiring them to update and to keep the three lastly presented letters in memory. The letters were presented for 1 s, with an ISI of 0.5 s.
When prompted, which could be at any given moment, participants provided their response by typing in three letters using the keyboard. If they failed to remember the correct letter, they provided a guessing-based answer. Four practice trials were completed by all Mean 3.2 ± 2.7 Note: Age, sex, education, and socioeconomic factors across decades (means ± standard deviation; frequencies).
The number-updating task had a columnized numerical 3-back design. Three boxes were present on the screen throughout the task, in which a single digit (1-9) was presented one at a time, from left to right during 1.5 s with an ISI of 0.5 s. During this ongoing sequence, participants had to judge whether the number currently presented in a specific box matched the last number presented in the same box (appearing three numbers before). For each presented number they responded yes/no by pressing one of two assigned keys ("yes" = right index finger; "no" = left index finger). Four test trials, each consisting of 30 numbers, followed after two practice trials. Performance was defined as the sum of correct responses across the four test trials, after discarding responses to the first three numbers in every trial (as these were not preceded by any numbers to be matched with). The maximum score was 108 (27 numbers × 4 trials). Reliability of this measure was estimated as 0.95 (Cronbach's alpha, n = 179).
In the spatial-updating task, three 3 × 3 square grids were presented next to each other on the computer screen. At the beginning of each trial, a blue circular object was, at the beginning of each trial, displayed at a random location within each grid. Participants were required to judge whether two items presented next to each other on the screen were identical or not, and provided yes/no responses by pressing assigned keys on the keyboard. The instructions were to respond as correctly and fast as possible.
In letter comparison, items consisted of two strings of four letters (a-z), constructed as to not constitute real words. Two strings were identical if they included the same letters in the same sequence, and non-identical when one letter differed between them. An item pair was presented until a response was provided, or for a maximum of 5 s (timeout

| Implicit learning
Implicit learning was operationalized as sequence learning and tested using the serial-reaction time test (SRTT: Nissen & Bullemer, 1987;Rieckmann et al., 2010;Seger, 1994). This task presents participants with sequences of events that they have to respond to, and measures the difference in reaction time between repeated and new sequences, considered to reflect implicit sequence learning. In this task, four squares were presented next to each other in a row on the computer screen. The squares were grouped together in two pairs (one more to the left and the other more to the right of the center of the screen). were to place the index finger of the hand being tested on a given computer key and, when cued, tap as fast as possible for the duration of the trial which was 25 s.

| Blood sampling
All participants provided a blood sample at the second testing session. These samples are stored at the Department of Biobank Research at Norrlands University Hospital in Umeå, Sweden.
Samples were collected for analyses of genetic and metabolomic factors potentially contributing to individual differences in cognition, brain measures, and age-related changes therein. Fasting was not required. A total of 40 ml blood was obtained, in equal amounts in four separate tubes, using a 1.3 mm diameter cannula. The time of blood sampling was registered and the samples were transported to the biobank for storage (at −80°C), maximum 2 hr after collection. Extraction of DNA will be performed for genotyping of single nucleotide polymorphisms of genes coding for DA genes (e.g., DA D1/2, COMT) as well as their methylation profiles to assess epigenetic factors. Other genetic polymorphisms related to brain integrity and cognition will also be assessed (e.g., APOE).

Structural MRI
High-resolution anatomical T1-weighted images were collected using a 3D fast spoiled gradient-echo sequence. Imaging parameters were as follows: 176 sagittal slices, thickness = 1 mm, repetition time

Cerebral perfusion
In order to assess cerebral perfusion, images were sampled using   The category "other" includes, among others, the following conditions: benign prostatic hyperplasia, chronic obstructive pulmonary disease, depression and/or anxiety, inflammation, movement disorders, musculoskeletal pain or substance abuse/dependence. was approximately 4.5 min, with a labeling time of 1.5 s, postlabeling delay time of 2 s, FOV = 240 × 240 mm, slice thickness of 4 mm, and an acquisition resolution of 8 × 512 (8 arms with 512 data points) with the number of averages set at 3. This sequence provided whole-brain perfusion in ml/100 g/min.

Functional MRI
Whole-brain functional images were acquired during three condi-

| Processing of brain imaging data
The preprocessing and analysis of selected measures of MRI and PET data are described in this section. The measures include the anatomical T1-weighted images, resting-state fMRI data, the [ 11 C] SCH23390 and [ 11 C]raclopride PET data.

| Volumetric assessments
Anatomical T1-weighted images were used to delineate subcortical structures with the Freesurfer 6.0 software (https://surfer.nmr.mgh. Functional images underwent slice-timing and movement correction, followed by distortion correction using subject-specific field maps. Three participants were excluded from the distortion correction procedure due to technical issues during field-map acquisition. Structural and functional data were subsequently co-registered and normalized using a study-specific template by Diffeomorphic Anatomical Registration using Exponentiated Lie algebra (DARTEL: Ashburner, 2007). Four individuals were excluded in the templategeneration step due to non-pathological anatomical deviations.
Following DARTEL, subject-specific flow fields were used to normalize images to MNI space, and images were subsequently smoothed with a 6-mm Gaussian kernel.
Additional preprocessing steps were completed to reduce spurious variance from non-neuronal sources: (i) demeaning and detrending each run, (ii) defining a multiple regression model including several nuisance regressors described below, (iii) finally, nuisance regression as setup in the previous step and temporal high-pass frequency filtering (threshold of 0.009 Hz) were applied simultaneously to not re-introduce nuisance signals (Hallquist et al., 2013). The nuisance regressors included mean cerebrospinal and white matter signal, Friston's 24-parameter motion model (six motion parameters, their squares and temporal derivatives ;Friston et al., 1996), and a binary vector of motion-contaminated volumes identified by the degree of frame wise displacement (FD). An FD metric that is independent of the definition of the center of rotation was used based on the transformation matrix instead of rotation directly (Jenkinson et al., 2002). Volumes with FD > 0.2 mm were flagged as motion contaminated. Finally, physiological nuisance regressors were included to control for spurious effects of respiration and heart rate using the Matlab PhysIO Toolbox v.5.0 (Kasper et al., 2017). A RETRICOR model (Glover et al., 2000;Hutton et al., 2011) was employed using Fourier expansions for the estimated phases of cardiac pulsation (up to third-order harmonics), respiration (up to forth-order harmonics), and first-order cardio-respiratory interactions.

| PET data
PET data were processed for two separate purposes. First, a characterization of striatal D1DR and D2DR, for which data processing followed the same steps for [ 11 C]SCH23390 and [ 11 C]raclopride images. Striatal regions were selected for baseline characterization of D1DR and D2DR in the DyNAMiC sample based on previous findings linking structural, functional, and DA receptor integrity of these regions to age-sensitive cognitive domains (Bäckman et al., 2000Nyberg, 2017). Furthermore, given their rich DA innervation, striatal regions serve as a good point of reference in terms of D1DR and D2DR estimates, as well as across the DyNAMiC and COBRA (Nevalainen et al., 2015) data sets enabling comparisons and pooling of PET data. The second purpose was for estimation of D1DR in cortical regions corresponding to nodes in known functional brain systems (Power et al., 2011). These estimates were later used in assessments of associations between D1DR and functional connectivity across cortical regions.
For both purposes, binding potential relative to nondisplaceable binding in a reference region (BP ND ; Innis et al., 2007), was used as an estimate of receptor availability (i.e., D1DR; D2DR) in target regions, using the cerebellum as reference. PET data were corrected for head movement by using frame-to-frame image coregistration, and co-registered with T1-weighted MRI data with re-slicing to MR voxel size using Statistical Parametric Mapping (SPM12: Wellcome Trust Centre for Neuroimaging, http://www.fil. ion.ucl.ac.uk/spm/). For the striatal regions putamen and caudate nucleus, the simplified reference tissue model (SRTM) was used to model regional time-activity course (TAC) data (Lammertsma & Hume, 1996). Regional TAC data were adjusted for partial volume

| Associations between cortical functional connectivity and D1DR
The second aim of the current study was to provide an initial characterization of associations between the functional connectome and D1DR. To that end, we present analyses and initial results from ongoing work in our group. To characterize the functional connectome, we employed a graph theoretical approach to quantify resting-state functional connectivity in terms of ROI-wise nodal strength of 243 cortical regions that have shown to be in good agreement with known functional brain systems (Power et al., 2011 SCH23390 PET (n = 175). To avoid age-related bias in the distribution of positive edges, only edges found to be positive in at least half of the sample were considered. Nodal D1DR was defined as the estimates of cortical D1DR extracted from each functional ROI during preprocessing, averaged across participants. The same connectivity and D1DR estimates were also computed within each decade, and for three larger groupings of young (20-39 years, n = 57), middle-aged (40-59 years, n = 57), and older adults (60-79, n = 62).
Stepwise linear and quadratic modeling was then conducted to investigate the link between D1DR and nodal strength.

| Statistical power
Because DyNAMiC investigates individual differences in rates of changes in brain and cognitive measures, we estimated the power to detect individual changes in cognition (McArdle & Nesselroade, 1994 In the simulations, the power to detect longitudinal change was approximated to be at least 88% when at least 80% reliability was considered (the reliability of most cognitive tests in DyNAMiC was estimated to be around 0.8-0.95) and variance of change was considered to be at least 20% of the baseline variance (in a similar imaging sample, Betula: Nilsson et al., 2004), the variance of longitudinal change was around 42% of the variance in initial level.

| Sample characteristics
Demographic information from the DyNAMiC sample is presented in Table 1. The educational level of the sample was relatively high, with 58.4% of participants reporting university-level education. This is consistent with Umeå being one of Sweden's main university cities, similarly reflected in the educational levels in other Umeå-based study samples (Nevalainen et al., 2015;Nilsson et al., 1997). Only a small group of participants were unemployed (3.9% of the full sample), while a majority of the sample reported some form of employment (69.1%), and a majority of older individuals (> 60 years) had retired (74.6%). Although the average number of children across the full sample (1 ± 1.9) was lower than the national figure of 1.7 (Statistics Sweden, 2019, https://www.scb.se/en/), numbers observed for individuals >40 years were higher (means ranging between 1.8 and 2.3 across age groups).
An overview of health parameters is presented in Table 2.
Examination of medical information shows that 50% of participants reported using some form of medication. The most common treatment was for hypertension, reported by 15.7% of participants, with a majority of these participants (78.6%) belonging to the two oldest age groups (>60 years). The prevalence of hypertension medication was overall 34.9% in participants over the age of 60. The second most common medical treatment was for asthma (8% of the sample), followed by hyperlipidemia (7%), and cardiovascular disease (6%).
These observations indicate that treatments regulating cardiovascular disease and risk factors, such as high blood pressure and cholesterol, were most prevalent overall, with numbers driven by the older segment of the sample. Average BMI values (ranging from 24.3 to 26.7 across age groups) were within the normal to overweight span (19)(20)(21)(22)(23)(24)(25)(26)(27)(28)(29)(30). In total, 19.6% of participants reported consuming nicotine (smoking and/or using snus), with the largest number of smokers found in the youngest (n = 6) and oldest (n = 4) groups.

| Cognitive performance
Distributions of responses across cognitive tests are presented in Figure 3a. Results of normality tests are presented in the Supporting Information. The distribution of scores from the episodic numberword recall test indicated that this task was difficult for participants to perform. The working memory number-updating task, on the other hand, had a high proportion of high scores from younger individuals (see Figure 3b), whereas most of the low scores in this task came from the older subjects. Overall, lowest performances were observed in older participants, except for the vocabulary and implicit learning tests (Figure 3b).
3.2.1 | The factor structure of episodic memory, working memory, and perceptual speed Given relatively different distributions of scores across cognitive tests, it is reasonable to expect that these tests might differentially represent their respective cognitive domains.
To validate the contribution of the considered cognitive domains to the observed data, the structure of the episodic memory, working memory, and perceptual speed was assessed through structural equation modeling (SEM). We investigated a similar latent structure to the one reported in the COBRA study that had the same test battery (Nevalainen et al., 2015) to relate the latent structure in our age-heterogeneous sample to the latent structure in COBRA's agehomogeneous sample. Out of 180 subjects, one participant had the spatial-updating task score missing, and one subject had the scores for all working memory tasks missing. Before fitting a SEM model, we excluded univariate outliers, that is the observations with the absolute value of the standardized score greater than 3.29 (the probability to observe such values when the data are normally distributed is <0.001). In total, five such univariate outliers were detected: two observations for number-word recall, two observations for num- to the COBRA study (Nevalainen et al., 2015), that used the same test battery to investigate cognition within a limited age span (64-68 years), the correlation between working and episodic memory cognitive domains was the strongest. replicating the rank order of BP ND s reported in autopsy work (Hall et al., 1994).

| Dopamine D2DR
Distributions of regional [ 11 C]raclopride BP ND estimates (PVE corrected) are presented in Figure 5c. The small size of the subsample exposed to [ 11 C]raclopride-PET imaging did not allow statistical evaluation of the distributions. Rank order of regional BP ND s was, however, in good concordance with earlier studies (Hall et al., 1994;Papenberg et al., 2019). Highest average BP ND s were observed in the putamen (4.17 ± 0.51), followed by the caudate nucleus (3.20 ± 0.43).

| Association of D1DR and functional connectivity across cortical regions
The distribution of regional D1DR across the cortex was found

| DISCUSS ION
This paper introduced the DyNAMiC study, designed to meet the current paucity of longitudinal multimodal data necessary for a comprehensive understanding of age-related changes in cognition across the adult lifespan. It is well documented that cross-sectional and longitudinal estimates of age-related changes in brain integrity and cognition deviate (Fjell et al., 2015;Nyberg et al., 2010;Raz et al., 2005;Rönnlund et al., 2005;Salthouse, 2010). Critically, only a few studies have explored longitudinal changes in the functional connectome (Chong et al., 2019;Malagurski et al., 2020;Ng et al., 2016;Pedersen et al., 2021),   (Persson et al., 2002); ~50% in 60-year olds (Carlsson et al., 2008); and 33% in adults 64-68 years, also recruited from Umeå (Nevalainen et al., 2015). The level of education was higher in DyNAMiC To create multifaceted measures of episodic memory, working memory, and perceptual speed, each domain was tested using three tests, including verbal, numerical, and figural materials, respectively.
The low scores on episodic number-word recall suggest that this test was difficult for participants to perform, consistent with results from a previous study using the same test battery (Nevalainen et al., 2015).
In contrast, a large proportion of participants (age < 55 years) displayed high scores on the working memory number-updating test, indicating that this was not a very challenging task. Given these observations, we assessed the factor structure of the episodic memory, working memory, and perceptual speed tests through SEM, evaluating the contribution of each subtest to its corresponding domain.

F I G U R E 6
Associations between cortical D1DR and functional nodal strength. (a) Surface projection of average cortical D1DR in 32k MSMAll HCP surface space (Glasser et al., 2016) and cortical nodes (color coded by nodal strength) in the Power atlas (Power et al., 2011 Modeling suggested the existence of three latent cognitive domains demonstrating shared variance across subtests (Figure 4). Given that SEM capitalizes on the shared variance across subtests, it is conceivable that using factor scores from such a model provides a good alternative to mean-based composite measures of performance.
Importantly, significant negative associations with age were evident for all three cognitive domains, in line with previous literature indicating their age sensitivity (Rönnlund et al., 2005;Salthouse, 2010).
In sum, modeling of cognitive performance indicated that the first wave of DyNAMiC provides the means to examine diverse and complex aspects of cognition in aging. Moreover, power analyses suggested that the power to detect longitudinal changes in cognition from DyNAMiC data is expected to be high.
DA modulation of synaptic activity enhances specificity in neuronal signal (Seamans & Yang, 2004;Shafiei et al., 2019;Shine et al., 2019), and past studies have reported associations of D1DR with brain activation and FC in young adults (Roffman et al., 2016;Turner et al., 2020). As such, age-related DA decline might constitute a basis for changes in the functional connectome. Effects of DA on FC within large-scale brain networks are, however, reported as diverse (Cole et al., 2013;Wallace et al., 2011), although previous studies indicate regional variability in the association between DA and FC (Tang et al., 2019;Xu et al., 2016), and that the spatial distribution of neurotransmitter receptors contribute to the brain's functional architecture (Dukart et al., 2021;Hansen et al., 2021).
This may have important implications for the functional connectome in aging, as indicated by a recent study showing that regional variability in age-related effects on FC was related to D1DR (Garzón et al., 2021). However, given that findings on regional variability in the association between D1DR and FC at a systems level are  (Cools & D'Esposito, 2011;Zahrt et al., 1997), it is important to note that our observation is not based on interindividual differences, but rather D1DR and FC across different regions. As such, further examination is needed to explore this non-linear D1DR-FC association across cortical network nodes in relation to individual differences in these measures. Given previous cross-sectional observations of age-related D1DR decline (Karrer et al., 2017), healthy aging might be accompanied by levels of D1DR occupancy outside the optimal range for efficient neural signaling, similar to other conditions characterized by D1DR deficiency (e.g., Parkinson's disease; Goldman-Rakic et al., 2000). This, in turn, may result in impaired cognitive function Lindenberger et al., 2008). Relatedly, degeneration within the mesocorticolimbic DA system has been implicated in cognitive decline and disease progression in Alzheimer's disease (Martorana & Koch, 2014;Trillo et al., 2013). Taken together, aberrant DA-FC associations may as such serve as a potential marker of cognitive decline in aging, and of older individuals at risk of converting to pathological aging.
Although SEM conveyed significant negative associations between cognitive domains and age, it is important to note that the current study on baseline data cannot characterize inter-individual differences in age-related trajectories of cognitive and brain measures. Instead, the main aim was to provide a comprehensive descriptive characterization of DyNAMiC baseline data, in parallel to assessing potential links between select core measures, which can be further explored in upcoming investigations. As such, whereas the stratification of participants into age groups does not capture the full extent and qualities of possible age-related effects, it reflects the lifespan design of the DyNAMiC sample. Due to the strict recruitment criteria, older DyNAMiC participants are considerably healthy, making it likely that a proportion of those elderly individuals are so called super agers (Harrison et al., 2012;Pudas et al., 2013;Yu et al., 2020), who throughout aging beyond 80 years display characteristics on par with those observed in 50-60 year olds . This may, already at baseline, be reflected in higher than expected levels of cognitive function and brain integrity, in turn attenuating effects of age. For instance, given that episodic memory is a highly age-sensitive domain (Gorbach et al., 2017;Rönnlund et al., 2005;Schaie, 1994), initial observations of small differences between age groups might indeed suggest an impact of the older participants' good health status.
Some challenges are associated with the design of DyNAMiC.
For instance, effects of attrition risk biasing longitudinal estimates of change (Eisner et al., 2019;Goodman & Blum, 1996;Lewin et al., 2018), but are an almost inevitable feature of longitudinal studies due to factors such as relocation, mortality, and arising MRI incompatibility. Attrition is often meaningful and non-ignorable, given significant associations of drop-out status and decline in brain and cognitive integrity (Josefsson et al., 2012;Nyberg & Pudas, 2019). Even though careful means were taken to achieve a healthy sample, it is indeed possible that some participants will convert from normal to pathological aging over time, which might affect several aspects of the brain-for instance the functional connectome (Filippi et al., 2020;Fox & Greicius, 2010;Sheline & Raichle, 2013;Zhang & Raichle, 2010). Accounting for dependencies between attrition and variables of interest will therefore be important in identifying reliable change-change associations between brain integrity and cognition (Gorbach et al., 2017;Josefsson et al., 2012;Little, 1995).
Finally, DyNAMiC does not include the oldest-old individuals (>80 years), in contrast to other longitudinal and multimodal brain imaging initiatives like the Human Connectome Project in Aging (Bookheimer et al., 2019), and the Umeå-based Betula study (Nilsson et al., 1997(Nilsson et al., , 2004. This demographic is in Sweden expected to increase by 50% between 2018 and 2028 (Statistics Sweden, 2019, https://www.scb.se/en/), but remains an age segment left out of most imaging studies to date. At the second time point, DyNAMiC will, however, be able to include returnees over the age of 80, providing information on potential changes during these individuals' transition into the oldest-old demographic.

| CON CLUS IONS
The first wave of DyNAMiC has provided a large multimodal data set, which will advance our understanding of lifespan alterations in human brain structure, function, and DA neurotransmission, as related to each other and to cognitive decline in aging. DyNAMiC is the largest DA D1 study worldwide, and will be able to examine trajectories and rates of change; identify onsets of brain and cognitive decline informing the optimal time point for interventions; tease apart shared and unique contributions of different brain measures to changes in cognition; and identify associations between functional and molecular brain systems as potential mechanisms of cognitive decline. Initial observations indicated that spatial configuration of functional regions depends on underlying DA receptor distribution, and for the first time revealed a non-linear effect of D1DR at a global neuronal level across the adult lifespan.

D ECL A R ATI O N O F TR A N S PA R EN C Y
The authors, reviewers and editors affirm that in accordance to the policies set by the Journal of Neuroscience Research, this manuscript presents an accurate and transparent account of the study being reported and that all critical details describing the methods and results are present.

ACK N OWLED G M ENTS
We thank the staff of the DyNAMiC project, Frida Magnusson and Emma Simonsson, staff at MRI and PET labs at Umeå University Hospital, and all our participants.

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest.

E TH I C S A PPROVA L
This study was approved by the Regional Ethical board and the local Radiation Safety Committee in Umeå, Sweden.

PEER R E V I E W
The peer review history for this article is available at https://publo ns.com/publo n/10.1002/jnr.25039.

DATA AVA I L A B I L I T Y S TAT E M E N T
Data from the DyNAMiC project cannot be made publicly available due to ethical and legal restrictions. However, access to these original data may be available upon request from the corresponding author.