Children s Hospital of Philadelphia Annual Progress Report: 2010 Formula Grant

Transcription

1 Children s Hospital of Philadelphia Annual Progress Report: 2010 Formula Grant Reporting Period July 1, 2012 June 30, 2013 Formula Grant Overview The Children s Hospital of Philadelphia received $3,548,977 for grant award period January 1, 2011 to December 31, Accomplishments for the reporting period are described below. Research Project 1: Project Title and Purpose Magnetic Resonance Imaging and Neurocognitive Assessment in Chronic Kidney Disease - Utilizing state of the art neuroimaging technology, including standard magnetic resonance imaging (MRI), arterial spin labeling (ASL), resting state functional MRI (fmri), and a concurrent matched control group in a cross-sectional study of 180 individuals, aged 8-25 years, we will test the hypothesis that subclinical vascular disease affects cognitive function in chronic kidney disease (CKD). Using novel pattern recognition methods, we will integrate these measures to develop a multi-parametric imaging phenotype in CKD that can identify high-risk groups. We hypothesize that level of kidney function, hypertension (HTN) and anemia affect resting cerebral blood flow (CBF) and functional performance. Achievement of our aims will enable targeted interventions to prevent and treat the cognitive deficits associated with CKD. Anticipated Duration of Project 1/1/ /31/2014 Project Overview To identify the causal mechanisms underlying the cognitive decline seen in CKD, we propose to use state of the art multimodal MRI and cognitive testing in a cross-sectional study of 90 subjects with Stage III, IV and V CKD (estimated GFR <90 ml/min/1.73m2), aged 8 25 years, and 90 controls matched on age and socioeconomic status, using insurance status as a proxy. Our specific aims are: 1. To demonstrate differences in brain imaging between the group of individuals with CKD and age-matched controls, manifested by a higher prevalence of subcortical hyperintensities on T2- weighted (FLAIR) MRI and reduction in ASL signal in subjects with CKD. We hypothesize that HTN and anemia in CKD cause alterations in CBF measured using ASL perfusion MRI, as well as hypoxic-ischemic injury in the brain, which can be measured via multi-parametric MRI (combination of FLAIR, T2, Proton density and T1-weighted images). Children s Hospital of Philadelphia 2010 Formula Grant Page 1

2 2. To explore the impact of CKD on structural brain integrity as measured by anatomic MRI with quantitative volumetric measurements with a particular focus on prefrontal cortex and frontotemporal regions. We hypothesize that evidence of frontotemporal brain injury will be correlated with deficits in executive function in subjects with kidney disease. 3. To characterize the impact of longer duration and increased severity of CKD, HTN and anemia on neurocognitive abilities, and correlate these with fmri findings. We hypothesize that cognitive deficits will be correlated with changes in regional CBF and/or connectivity within specific brain networks supporting these neurocognitive functions. 4. To develop individual-patient biomarkers by combining structural, ASL and fmri data in multi-parametric classification. Our goal is to ultimately find imaging patterns that in future studies can predict high risk of neurocognitive decline and thereby identify subjects for targeted interventions. Principal Investigator Susan L. Furth, MD, PhD Professor, Pediatrics Children's Hospital of Philadelphia 3615 Civic Center Boulevard Philadelphia, PA Other Participating Researchers Abbas Jawad, PhD; Divya Moodalbail, MD; Kevin Meyers, MD; Madhura Pradhan, MD; Jerilynn Radcliffe, PhD; Rebecca Ruebner, MD; Robert Schultz, PhD; Karuna Shekdar, MD - employed by The Children s Hospital of Philadelphia Christos Davatzikos, PhD; John Detre, PhD; Ruben Gur, PhD - employed by University of Pennsylvania Stephen Hooper, PhD employed by University of North Carolina Expected Research Outcomes and Benefits We will identify differences in brain structure and functional performance between CKD and control group children and young adults. Structural MRI (smri) will assess the regional distribution of grey matter, white matter and cerebrospinal fluid (CSF), which will determine whether there are correlations between CKD related clinical variables, brain atrophy and cognitive performance, and whether or not these relationships are spatially heterogeneous. Structural MRI will provide measurements of the regional distribution of cerebrovascular disease (CVD), as HTN is exceedingly common in CKD and is one of the highest risk factors for CVD. Additionally, ASL MRI will provide measurements of regional blood perfusion, which will provide mechanistic insights into the pathogenesis of observed structural changes based on correlations between CKD-related clinical variables, brain atrophy and cognitive performance, and whether or not these relationships are spatially heterogeneous. Using both ASL MRI and resting state fmri we will examine brain activity in the absence of external stimuli and this will Children s Hospital of Philadelphia 2010 Formula Grant Page 2

3 allow us to evaluate functional changes in these patients that correlate with clinical and cognitive variables. State of the art pattern recognition methods will allow us to integrate all these types of measurements and develop a multi-parametric imaging phenotype in CKD. This phenotype can be used to predict subsequent cognitive decline in future longitudinal studies so that interventions can be implemented to improve functional performance in this population. We are studying children and young adults with CKD, as this will allow us to identify neurocognitive and neuroanatomic changes due to kidney disease in the absence of underlying CVD, which is ubiquitous in adults with CKD. The long term benefit of this project will be improvement in the health status of the 1 in 9 Americans with CKD, as cognitive impairment may affect adherence to the complex medical regimens which these patients are routinely prescribed, and may contribute substantially to decreasing the adverse outcomes and substantial cost associated with end stage renal disease (>$36 billion in 2007). Summary of Research Completed Methods and Results: Since the prior report, subjects have enrolled in the study, bringing the total enrollees to 51 Chronic Kidney Disease (CKD) subjects and 40 Controls, as of May15, This total does not include all enrollees, as some enrolled participants have not yet completed all testing. A pilot longitudinal arm was added to the study to retest CKD participants that have: begun or stopped dialysis, received a transplant, or had 30% change in kidney function by either estimating formula since their initial study visit. An estimated 10% (n = 10) of CKD participants will qualify for inclusion in the longitudinal arm, along with matched controls. CHOP s IRB has just approved the addition of this longitudinal arm, however enrollment has not yet begun. Demographics and socioeconomic status: Table 1 presents demographic data on 43 participants with CKD and 31 Controls. The recruited number of Control participants deliberately lags behind CKD participants so that the researchers can tailor control enrollment strategies to appropriately match on age, gender, and socioeconomic status (using insurance status as a proxy). Control recruitment strategies are frequently reassessed and altered to improve matching. Currently, CKD cohort is 65% male, 26% black, Control cohort is 42% male, 65% black. Maternal/paternal education and socioeconomic status between CKD and controls is similar, but the researchers are actively recruiting controls from different geographic areas to more closely match on SES measures. Age, gender, race, ethnicity, maternal/paternal education and household income data are also presented in Table 1. Lab Data: Lab values are presented in Table 2 and show expected significant differences in kidney function in Blood Urea Nitrogen and Serum Creatinine in CKD subjects vs controls. Ambulatory Blood Pressure Monitor Data: Blood pressure data are presented on 31 CKD and 19 Controls in Table 3. Compared to controls, CKD cohort showed higher blood pressure readings in every parameter, including reduced nocturnal dip. This is consistent with clinical patterns seen in children with CKD. Children s Hospital of Philadelphia 2010 Formula Grant Page 3

4 Traditional Neurocognitive Battery: 41 CKD and 26 Controls completed the traditional neurocognitive battery, which consisted of the Wechsler Abbreviated Scales of Intelligence, Conners' Continuous Performance Test II, Delis-Kaplan Executive Function System Tower Subtest, Wechsler Intelligence Scale for Children Fourth Edition Integrated Digit and Spatial Span Tasks (for participants 16 years old) or the Wechsler Memory Scale Third Edition Digit and Spatial Span Tasks (for participants 17 years old). Individual mean Z-scores of sub-tests from each battery were combined and converted into constructs representing language, attention, inhibitory control, planned problem solving, set shift, visuospatial processing, verbal memory, and visual working memory. CKD participants showed weaknesses most notably in the areas of attention and verbal memory with scores more than 0.5 SD different from controls (Table 4). Results from the parent report BRIEF (for participants 8-18 years old) and BRIEF Adult form (self-report for participants 18 years old) are presented in Table 5. Domains where CKD participants had the most salient weaknesses compared to controls were in the self and parent completed forms for Plan/Organize in the Metacognition Index and in the parent-completed form for the Global Executive Composite. Computerized Neurocognitive Battery: 42 CKD and 25 Controls completed the Computerized Neurocognitive Battery, comprised of Speed and Accuracy performance measures in the following domains: Abstraction/Mental Flexibility, Attention, Working Memory, Verbal Memory, Facial Memory, Spatial Memory, Language, Non-Verbal Reasoning, Spatial Orientation, Emotion Identification, Emotion Differentiation, Age Differentiation, Motor Speed, and Sensorimotor Speed. For the purpose of this comparison, Controls were normalized to a Z- score of zero, as presented in Fig 1. Domains where CKD participants performed worse, with an effect size of 0.5 SD or greater, were Attention, Facial Memory, Non-verbal Reasoning, Spatial Orientation, and Emotion Identification in Accuracy, and Spatial Memory in Speed. Neuroimaging Data: Arterial Spin Labeling (ASL) MRI analysis of cerebral blood flow (CBF) was completed for 39 CKD and 19 Controls. Comparisons between groups (CKD vs. Control subjects) and gender (male vs. female) were applied in measurements of CBF, Hematocrit (Hct) and age using Wilcoxon rank test. Univariate associations between all measurements of CBF, Hct and age were analyzed by using Spearman correlations. Multiple stepwise regression models were used to predict CBF outcomes using group, age and Hct. CKD subjects showed a trend of higher gray matter (GM) and global CBF. There was no statistical significant gender difference in CBF. Spearman correlation analyses showed significant univariate associations between five outcomes among the controls (Table 6). CKD showed significant but weaker associations between GM, global CBF, Hct and age but not between WM CBF and age. Figure 2 shows a scatter plot of age and WM CBF for CKD and control subjects. For controls there is a significant correlation between age and WM CBF, but in older CKD patients WM CBF appears to be increased. Multiple regression analysis indicated that Hct and age are the dominant predictors of CBF. There was no significant effect of group status in the multiple regression analysis. Structural MRI (smri) processing pipeline was applied on the image data of 60 subjects (40 CKD and 20 controls).. A completely automated pipeline (with the exception of a minimal manual correction on automatically generated brain masks on T1-weighted images) was developed and applied for processing structural MR images. The processing pipeline modules Children s Hospital of Philadelphia 2010 Formula Grant Page 4

5 were described in detail in the 2011 Progress Report. While keeping the general pipeline unchanged, the researchers improved it by replacing two main modules by new, state-of-the-art methods developed in-house: 1) A new multi-atlas label fusion method, using DRAMMS registration to align atlas images to subject space, was developed for segmenting the brain into a set of anatomical regions of interest (ROIs). Compared to the previous single-atlas based approach, the label-fusion approach significantly improved the accuracy of the segmentation into ROIs. Furthermore, a hierarchical structure was provided with the new set of ROIs, which allowed the researchers to group them into larger anatomical regions (for example brain lobes) (Figure 3). Volumetric measurements for normal and abnormal (with lesion) tissue, and mean FA and CBF values were calculated both for single and derived ROIs; 2) Regional pattern analysis on tissue density maps was performed using a new optimally-discriminative voxel-based analysis framework (ODVBA) (Figure 4). Similar to classical voxel based morphometry (VBM) approaches, ODVBA calculates a statistical parametric map of group differences between CKD and Controls. However, instead of smoothing images with a Gaussian filter prior to analysis, ODVBA determines the optimal spatially adaptive smoothing of images (i.e. in the optimally discriminative direction) using machine learning techniques. Finally, a new machine learning methodology, the brain development index (BDI), which summarizes the multivariate pattern of structural brain development, was applied on the imaging data to compare group differences in brain development between normal controls and CKD patients. Functional MRI: The researchers have begun to integrate the fmri resting state images with neurocognitive testing data. One of the primary approaches is to compute average correlations across time between each intracerebral voxel with every other voxel, yielding an overall connectivity score that can be integrated with phenotypic data (i.e., neuropsychological testing). The researchers are currently examining associations between prefrontal cortex regions and performance tasks, specifically for verbal memory, attention, and BRIEF working memory. Preliminary analyses suggest individuals with poorer BRIEF working memory scores, exhibit poorer connectivity between portions of dorsolateral prefrontal cortex and the rest of the brain (rvalues for these correlations exceed 0.4). Further analyses in these areas are proceeding. Publications and Abstracts: Neuroimaging in Chronic Kidney Disease: Review of Structural and Functional Findings in Children and Adults- mini-review published on CJASN website May 30, Arterial Spin Labeled Perfusion MRI in Children with Chronic Kidney Disease Poster presentation at the April 2013 International Society for Magnetic Resonance in Medicine. Neuropsychological Profile of Children and Young Adults with Moderate to Severe Chronic Kidney Disease Poster presentation at the January 2013 International Conference on Dialysis and published January 2013 in the special online edition of 'Blood Purification Journal'. Milestones: During this period, 3 of 3 milestones were met: 1) Recruit 50 subjects, schedule study visits: 51 CKD subjects and 40 Controls have scheduled study visits. 2) Review data obtained on initially recruited participants at steering committee meeting: Demographics, lab, blood pressure, neurocognitive, and neuroimaging data were reviewed at Children s Hospital of Philadelphia 2010 Formula Grant Page 5

6 quarterly in-person steering committee meetings. Additionally, biweekly conference calls were held to review progress, address any pertinent issues, and guide course of action. 3) Modify protocol if needed and submit renewals/modifications to IRB: During this reporting period, the protocol was amended to add DuPont, Jefferson, and University of Pennsylvania Hospitals as recruitment sites, to include post-card mailers as a recruitment strategy, and to add a pilot longitudinal arm to retest a small number of participants that have had a significant change in kidney function since their initial study visit, along with matched controls. Table 1: Demographics and Socioeconomic Status CKD (n = 43) Control (n = 31) Gender Male 65% 42% Race Black White Asian 26% 67% 7% 65% 32% 3% Ethnicity Hispanic 5% 10% Age 8-16 yrs 74% 71% Mother s Education (years) Father s Education (years) Household Income Per Year (self report) yrs > >16 < $6,000 $6,001 - $18,000 $18,001 - $30,000 $30,001 - $75,000 > $75,000 26% 37% 51% 12% 49% 35% 16% 10% 4% 10% 30% 48% 29% 41% 36% 23% 48% 38% 14% 17% 0% 17% 43% 23% Table 2: Lab Values CKD n = 43 Sodium Potassium Chloride Carbon Dioxide Urea Nitrogen Serum Creatinine Glucose Calcium CKD (n = 43) Control (n = 31) Mean ± SD Mean ± SD ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± 0.32 Children s Hospital of Philadelphia 2010 Formula Grant Page 6

7 Table 3: Ambulatory Blood Pressure Monitor Data CKD (n = 31) Controls (n = 19) Systolic Diastolic Systolic Diastolic Load * Index + Load * Index + Load * Index + Load * Index + (%) (%) (%) (%) Waking 28 ± ± ± ± ± ± ± 0.7 Sleeping 25 ± ± ± ± 9 12 ± Total 28 ± ± ± ± ± 7 11 ± ± 0.7 Dipping 11% ± 6% 16% ± 7% 12% ± 7% 19% ± 9% *Load: Percentage of readings > 95 th %ile, Normal <25%; Intermediate 25 50%; Abnormal >50% + Index: Mean ambulatory BP, an index > 1.0 is considered hypertensive Reference values for Dipping: Normal > 10% Table 4: Traditional Neurocognitive Battery Mean Z-Scores ± Standard Deviation Domains CKD (n = 41) Control (n = 26) Language Attention Inhibitory Control Problem Solving Set Shifting Visuo-spatial skills Verbal memory Visual Work Memory ± ± ± ± ± ± ± ± ± ± ± 0.12 ± ± ± ± ± 1.11 Table 5: BRIEF scores Mean t-scores ± Standard Deviation BRIEF Adult Form CKD Control (n = 8) (n = 6) Behavioral Regulation Index Inhibit Shift Emotional Control Self-Monitor Metacognition Index Initiate Working Memory Plan/Organize Task Monitor Organization of Materials Global Executive Composite ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± BRIEF Parent Form CKD Control (n = 33) (n = 20) ± ± ± ± (n/a) ± ± ± ± ± ± ± ± ± ± ± (n/a) ± ± ± ± ± ± ± Children s Hospital of Philadelphia 2010 Formula Grant Page 7

8 Figure 1: Results of Computerized Neurocognitive Battery (Means+/-SEM of the CKD group compared to controls who served as the normative group) Table 6: Spearman correlations for univariate associations between CBF, age and Hct in control (blue font) and CKD subjects (red font). Data are presented as correlation coefficients and P values. *P<0.05. Control subjects show strong correlations for all metrics. No associations were found between WM CBF and age (Roh=-0.20, P=0.24) or between Hct and age (Roh=0.27, P=0.09) for CKD subjects. Figure 2: Scatter plots showing correlations between age and white matter CBF in control (blue circle) and CKD (red star) subjects. A significant correlation between age and white matter CBF was found in control (Roh=-0.55, P=0.01) but not in CKD subjects (Roh=-0.20, P=0.24). Children s Hospital of Philadelphia 2010 Formula Grant Page 8

9 Figure 3: ROI segmentation of one of NiCK subjects (nick001). The ROIs are color-coded for a better visualization. a) Single ROI s; b) derived ROI s on the left hemisphere (frontal, temporal, parietal, occipital, limbic lobes and cerebellum) a b Figure 4: Preliminary results show significant regions of group-differences between the controls and CKDs. Voxelwise significance values are calculated using ODVBA and are corrected for multiple comparisons using FDR correction. Significant voxels (q<0.05) are overlaid in red on the template image. Children s Hospital of Philadelphia 2010 Formula Grant Page 9