Impact of inherent variability and experimental parameters on the reliability of small animal PET data

Deep-learning-derived input function in dynamic [18F]FDG PET imaging of mice

Dynamic positron emission tomography and kinetic modeling play a critical role in tracer development research using small animals. Kinetic modeling from dynamic PET imaging requires accurate knowledge of an input function, ideally determined through arterial blood sampling. Arterial cannulation in mice, however, requires complex, time-consuming and terminal surgery, meaning that longitudinal studies are impossible. The aim of the current work was to develop and evaluate a non-invasive, deep-learning-based prediction model (DLIF) that directly takes the PET data as input to predict a usable input function. We first trained and evaluated the DLIF model on 68 [18F]Fluorodeoxyglucose mouse scans with image-derived targets using cross validation. Subsequently, we evaluated the performance of a trained DLIF model on an external dataset consisting of 8 mouse scans where the input function was measured by continuous arterial blood sampling. The results showed that the predicted DLIF and image-derived targets were similar, and the net influx rate constants following from Patlak modeling using DLIF as input function were strongly correlated to the corresponding values obtained using the image-derived input function. There were somewhat larger discrepancies when evaluating the model on the external dataset, which could be attributed to systematic differences in the experimental setup between the two datasets. In conclusion, our non-invasive DLIF prediction method may be a viable alternative to arterial blood sampling in small animal [18F]FDG imaging. With further validation, DLIF could overcome the need for arterial cannulation and allow fully quantitative and longitudinal experiments in PET imaging studies of mice.

Investigation of Early Death-Induced Changes in Rat Brain by Solid Phase Microextraction via Untargeted High Resolution Mass Spectrometry: In Vivo versus Postmortem Comparative Study

Analysis of brain samples obtained postmortem remains a standard approach in neuroscience, despite often being suboptimal for inferring roles of small molecules in the pathophysiology of brain diseases. Sample collection and preservation further hinders conclusive interpretation of biomarker analysis in autopsy samples. We investigate purely death-induced changes affecting rat hippocampus in the first hour of postmortem interval (PMI) by means of untargeted liquid chromatography-mass spectrometry-based metabolomics. The unique possibility of sampling the same brain area of each animal both in vivo and postmortem was enabled by employing solid phase microextraction (SPME) probes. Four millimeter probes coated with mixed mode extraction phase were used to sample awake, freely roaming animals, with 2 more sampling events performed after death. Significant changes in brain neurochemistry were found to occur as soon as 30 min after death, further progressing with increasing PMI, evidenced by relative changes in levels of metabolites and lipids. These included species from several distinct groups, which can be classified as engaged in energy metabolism-related processes, signal transduction, neurotransmission, or inflammatory response. Additionally, we perform thorough analysis of interindividual variability in response to death, which provides insights into how this aspect can obscure conclusions drawn from an untargeted study at single metabolite and pathway level. The results suggest high demand for systematic studies examining the PMI time course with in vivo sampling as a starting point to eliminate artifacts in the form of neurochemical changes assumed to occur in vivo.

Evaluation of Precision in Optoacoustic Tomography for Preclinical Imaging in Living Subjects

Optoacoustic tomography (OT) is now widely used in preclinical imaging; however, the precision (repeatability and reproducibility) of OT has yet to be determined. Methods: We used a commercial small-animal OT system. Measurements in stable phantoms were used to independently assess the impact of system variables on precision (using coefficient of variation, COV), including acquisition wavelength, rotational position, and frame averaging. Variables due to animal handling and physiology, such as anatomic placement and anesthesia conditions, were then assessed in healthy nude mice using the left kidney and spleen as reference organs. Temporal variation was assessed by repeated measurements over hours and days both in phantoms and in vivo. Sensitivity to small-molecule dyes was determined in phantoms and in vivo; precision was assessed in vivo using IRDye800CW. Results: OT COV in a stable phantom was less than 2.8% across all wavelengths over 30 d. The factors with the greatest impact on signal repeatability in phantoms were rotational position and user experience, both of which still resulted in a COV of less than 4% at 700 nm. Anatomic region-of-interest size showed the highest variation, at 12% and 18% COV in the kidney and spleen, respectively; however, functional SO2 measurements based on a standard operating procedure showed an exceptional reproducibility of less than 4% COV. COV for repeated injections of IRDye800CW was 6.6%. Sources of variability for in vivo data included respiration rate, degree of user experience, and animal placement. Conclusion: Data acquired with our small-animal OT system were highly repeatable and reproducible across subjects and over time. Therefore, longitudinal OT studies may be performed with high confidence when our standard operating procedure is followed.

Comparative Analysis between [(18)F]Fludarabine-PET and [(18)F]FDG-PET in a Murine Model of Inflammation

Lymphoma research has advanced thanks to introduction of [(18)F]fludarabine, a positron-emitting tool. This novel radiotracer has been shown to display a great specificity for lymphoid tissues. However, in a benign process such as inflammation, the uptake of this tracer has not been questioned. Indeed, in inflammatory zones, elevated glucose metabolism rate may result in false-positives with [(18)F]FDG-PET Imaging. In the present investigation, it has been argued that cells, involved in inflammation, might be less avid of [(18)F]fludarabine. To generate inflammation, Swiss mice were intramuscularly injected with 0.1 mL of turpentine oil into the right front paw. Imaging sessions with (18)F-labeled tracers named above were conducted on days 5 and 25 after inoculation. For each animal, volumes of interest (VOI), delineating the muscle of the inflamed (IP) and normal paws (NP), were determined on PET scans. For characterization of inflammation, muscle samples from IP and NP were stained with hematoxylin and eosin (H&E). In early (day 5) inflammation, [(18)F]FDG accumulation was 4.00 ± 1.65 times greater in the IP than in the contralateral NP; for [(18)F]fludarabine, this IP/NP ratio was 1.31 ± 0.28, resulting in a significant difference between radiotracer groups (p < 0.01). In late (day 25) inflammation, the IP/NP ratios were 2.07 ± 0.49 and 1.03 ± 0.07, for [(18)F]FDG and [(18)F]fludarabine, respectively (p < 0.001). [(18)F]Fludarabine showed significantly weaker uptake in inflammation when compared with [(18)F]FDG. This encouraging finding suggests that [(18)F]fludarabine-PET might well be a robust approach for distinguishing tumor from inflammatory tissue, avoiding false-positive PET results and thus enabling an accurate imaging of lymphoma.

The Effects of Physiological and Methodological Determinants on 18F-FDG Mouse Brain Imaging Exemplified in a Double Transgenic Alzheimer Model

Introduction:

In this study, the influence of physiological determinants on 18F-fluoro-d-glucose (¹⁸F-FDG) brain uptake was evaluated in a mouse model of Alzheimer disease.

Materials and Methods:

TASTPM (Tg) and age-matched C57BL/6 J (WT) mice were fasted for 10 hours, while another group was fasted for 20 hours to evaluate the effect of fasting duration. The effect of repeatedly scanning was evaluated by scanning Tg and WT mice at days 1, 4, and 7. Brain ¹⁸F-FDG uptake was evaluated in the thalamus being the most indicative region. Finally, the cerebellum was tested as a reference region for the relative standard uptake value (rSUV).

Results:

When correcting the brain uptake for glucose, the effect of different fasting durations was attenuated and the anticipated hypometabolism in Tg mice was demonstrated. Also, with repeated scanning, the brain uptake values within a group and the hypometabolism of the Tg mice only remained stable over time when glucose correction was applied. Finally, hypometabolism was also observed in the cerebellum, yielding artificially higher rSUV values for Tg mice.

Conclusion:

Corrections for blood glucose levels have to be applied when semiquantifying ¹⁸F-FDG brain uptake in mouse models for AD. Potential reference regions for normalization should be thoroughly investigated to ensure that they are not pathologically affected also by afferent connections.

Quality of Animal Experiments in Anti-Angiogenic Cancer Drug Development--A Systematic Review

Translation from preclinical animal research to clinical bedside has proven to be difficult to impossible in many fields of research (e.g. acute stroke, ALS and HIV vaccination development) with oncology showing particularly low translation rates (5% vs. 20% for cardiovascular diseases). Several investigations on published preclinical animal research have revealed that apart from plain species differences, translational problems can arise from low study quality (e.g. study design) or non-representative experimental conditions (e.g. treatment schedule). This review assessed the published experimental circumstances and quality of anti-angiogenic cancer drug development in 232 in vivo studies. The quality of study design was often insufficient; at least the information published about the experiments was not satisfactory in most cases. There was no quality improvement over time, with the exception of conflict of interest statements. This increase presumably arose mainly because journal guidelines request such statements more often recently. Visual inspection of data and a cluster analysis confirmed a trend described in literature that low study quality tends to overestimate study outcome. It was also found that experimental outcome was more favorable when a potential drug was investigated as the main focus of a study, compared to drugs that were used as comparison interventions. We assume that this effect arises from the frequent neglect of blinding investigators towards treatment arms and refer to it as hypothesis bias. In conclusion, the reporting and presumably also the experimental performance of animal studies in drug development for oncology suffer from similar shortcomings as other fields of research (such as stroke or ALS). We consider it necessary to enforce experimental quality and reporting that corresponds to the level of clinical studies. It seems that only clear journal guidelines or guidelines from licensing authorities, where failure to fulfill prevents publication or experimental license, can help to improve this situation.

Assessment of the Effects of 6 Standard Rodent Diets on Binge-Like and Voluntary Ethanol Consumption in Male C57BL/6J Mice

BACKGROUND

In recent years, much attention has been given to the lack of reproducibility in biomedical research, particularly in preclinical animal studies. This is a problem that also plagues the alcohol research field, particularly in consistent consumption in animal models of alcohol use disorders. One often overlooked factor that could affect reproducibility is the maintenance diet used in preclinical studies.

METHODS

Herein, 2 well-established models of alcohol consumption, the "drinking in the dark" (DID) procedure and the continuous 2-bottle choice (C2BC) paradigm, were employed to determine the effects of diet on ethanol (EtOH) consumption. Male C57BL/6J mice were given 1 of 6 standard rodent chow diets obtained from Purina LabDiet(®) , Inc. (Prolab(®) RMH 3000) or Harlan(®) Laboratories, Inc. (Teklad Diets T.2916, T.2918, T.2920X, T.7912, or T.8940). A separate group of animals were used to test dietary effects on EtOH pharmacokinetics and behavioral measures following intraperitoneal (IP) injections of various doses of EtOH.

RESULTS

Mice eating Harlan diets T.2916 (H2916) and T.2920X (H2920) consumed significantly less EtOH and exhibited lower blood EtOH concentrations (BECs) during DID; however, during C2BC, animals maintained on Harlan T.7912 (H7912) consumed more EtOH and had a higher EtOH preference than the other diet groups. EtOH consumption levels did not stem from changes in alcohol pharmacokinetics, as a separate group of animals administered EtOH IP showed no difference in BECs. However, animals on Harlan diet T.2920X (H2920) were more sensitive to alcohol-induced locomotor activity in an open-field task. No diet-dependent differences were seen in alcohol-induced sedation as measured with loss of righting reflex.

CONCLUSIONS

Although these data do not identify a specific mechanism, together, they clearly show that the maintenance diet impacts EtOH consumption. It is incumbent upon the research community to consider the importance of describing nutritional information in methods, which may help decrease interlaboratory reproducibility issues.

Imaging dopaminergic dysfunction as a surrogate marker of neuropathology in a small-animal model of HIV

The dopaminergic system is especially vulnerable to the effects of human immunodeficiency virus (HIV) infection, rendering dopaminergic deficits early surrogate markers of HIV-associated neuropathology. We quantified dopamine D2/3 receptors in young HIV-1 transgenic (Tg) (n  =  6) and age-matched control rats (n  =  7) and adult Tg (n  =  5) and age-matched control rats (n  =  5) using [18F]fallypride positron emission tomography (PET). Regional uptake was quantified as binding potential (BPND) using the two-tissue reference model with the cerebellum as the reference. Time-activity curves were generated for the ventral striatum, dorsal striatum, thalamus, and cerebellum. Whereas BPND values were significantly lower in the ventral striatum (p < .001) and dorsal striatum (p  =  .001) in the adult Tg rats compared to controls rats, they were significantly lower only in the dorsal striatum (p < .05) in the young rats. Tg rats had smaller striatal volumes on magnetic resonance imaging. We also found lower expression levels of tyrosine hydroxylase on immunohistochemistry in the Tg animals. Our findings suggest that progressive striatal D2/3 receptor deficits occur in Tg rats as they age and can be detected using small-animal PET imaging. The effectiveness of various approaches in preventing or halting this dopaminergic loss in the Tg rat can thus be measured preclinically using [18F]fallypride PET as a molecular imaging biomarker of HIV-associated neuropathology.

Towards a reproducible protocol for repetitive and semi-quantitative rat brain imaging with (18) F-FDG: exemplified in a memantine pharmacological challenge

The standard uptake value (SUV), commonly used to quantify (18)F-FluoroDeoxyGlucose (FDG) uptake in small animal brain PET imaging, is affected by many factors. In this study the influence of fasting times, inter-scan duration and repetitive scanning on the variability of different SUV measures is investigated. Additionally it is demonstrated that these variables could adversely influence the outcome of a pharmacological challenge when not accounted for. Naive Sprague-Dawley rats (n=20) were randomly divided into five different fasting groups (no fasting up to 24h of fasting). SUV brain uptake values were reproducible in naive animals when a fasting period of at least 12h is used and for shorter fasting periods SUV values need to be corrected for the glucose level. Additionally, a separate animal group (n=6) was sufficiently fasted for 16h and in a longitudinal setting being scanned six times in three weeks. Especially with short inter-scan durations, increasing glucose levels were found over time which was attributed to increased stress due to repeated food deprivation, altered food intake or scan manipulations. As a result, even with controlled and sufficient fasting, blood glucose levels should be taken into account for data quantification. Strikingly, even the brain activation effects of an NMDA-antagonist challenge with memantine could not be detected in experiments with a short inter-scan duration if glucose levels were not taken into account. Correcting for glucose levels decreases the inter- and intra-animal variability for rat brain imaging. SUV corrected for glucose levels yields the lowest inter-animal variation. However, if the body weight changes significantly, as in a long experiment, quantification based on the glucose corrected percentage injected dose (and not SUV) is recommendable as this yields the lowest intra-animal variation.

FDG-PET imaging in mild traumatic brain injury: a critical review

Traumatic brain injury (TBI) affects an estimated 1.7 million people in the United States and is a contributing factor to one third of all injury related deaths annually. According to the CDC, approximately 75% of all reported TBIs are concussions or considered mild in form, although the number of unreported mild TBIs (mTBI) and patients not seeking medical attention is unknown. Currently, classification of mTBI or concussion is a clinical assessment since diagnostic imaging is typically inconclusive due to subtle, obscure, or absent changes in anatomical or physiological parameters measured using standard magnetic resonance (MR) or computed tomography (CT) imaging protocols. Molecular imaging techniques that examine functional processes within the brain, such as measurement of glucose uptake and metabolism using [¹⁸F]fluorodeoxyglucose and positron emission tomography (FDG-PET), have the ability to detect changes after mTBI. Recent technological improvements in the resolution of PET systems, the integration of PET with magnetic resonance imaging (MRI), and the availability of normal healthy human databases and commercial image analysis software contribute to the growing use of molecular imaging in basic science research and advances in clinical imaging. This review will discuss the technological considerations and limitations of FDG-PET, including differentiation between glucose uptake and glucose metabolism and the significance of these measurements. In addition, the current state of FDG-PET imaging in assessing mTBI in clinical and preclinical research will be considered. Finally, this review will provide insight into potential critical data elements and recommended standardization to improve the application of FDG-PET to mTBI research and clinical practice.

PET∕CT imaging evidence of FUS-mediated (18)F-FDG uptake changes in rat brain

PURPOSE

Transcranial focused ultrasound (FUS) delivers highly focused acoustic energy to a small region of the brain in a noninvasive manner. Recent studies have revealed that FUS, which is administered either in pulsed or continuous waves, can elicit or suppress neural tissue excitability. This neuromodulatory property of FUS has been demonstrated via direct motion detection, electrophysiological recordings, functional magnetic resonance imaging (fMRI), confocal imaging, and microdialysis sampling of neurotransmitters. This study presents new evidence of local increase in glucose metabolism induced by FUS to the rat brain using FDG (18-fludeoxyglucose) positron emission tomography (PET).

METHODS

Sprague-Dawley rats underwent sonication to a unilateral hemispheric area of the brain prior to PET scan. The pulsed sonication (350 kHz, tone burst duration of 0.5 ms, pulse repetition frequency of 1 kHz, and duration of 300 ms) was applied in 2 s intervals for 40 min immediately after the FDG injection via tail vein. Subsequently, the PET was acquired in dynamic list-mode to image FDG activity for an hour, and reconstructed into a single volume representing standardized uptake value (SUV). The raw SUV as well as its asymmetry index (AI) were measured from five different volume-of-interests (VOIs) of the brain for both hemispheres, and compared between sonicated and unsonicated groups.

RESULTS

Statistically significant hemispheric changes in SUV were observed only at the center of sonication focus within the FUS group [paired t-test; t(7) = 3.57, p < 0.05]. There were no significant hemispheric differences in SUV within the control group in any of the VOIs. A statistically significant elevation in AI (t-test; t(7) = 3.40, p < 0.05) was observed at the center of sonication focus (7.9 ± 2.5%, the deviations are in standard error) among the FUS group when compared to the control group (-0.8 ± 1.2%).

CONCLUSIONS

Spatially distinct increases in the glucose metabolic activity in the rat brain is present only at the center of sonication focus, suggesting localized functional neuromodulation mediated by the sonication.

FDG kinetic modeling in small rodent brain PET: optimization of data acquisition and analysis

BACKGROUND

Kinetic modeling of brain glucose metabolism in small rodents from positron emission tomography (PET) data using 2-deoxy-2-[(18) F]fluoro-d-glucose (FDG) has been highly inconsistent, due to different modeling parameter settings and underestimation of the impact of methodological flaws in experimentation. This article aims to contribute toward improved experimental standards. As solutions for arterial input function (IF) acquisition of satisfactory quality are becoming available for small rodents, reliable two-tissue compartment modeling and the determination of transport and phosphorylation rate constants of FDG in rodent brain are within reach.

METHODS

Data from mouse brain FDG PET with IFs determined with a coincidence counter on an arterio-venous shunt were analyzed with the two-tissue compartment model. We assessed the influence of several factors on the modeling results: the value for the fractional blood volume in tissue, precision of timing and calibration, smoothing of data, correction for blood cell uptake of FDG, and protocol for FDG administration. Kinetic modeling with experimental and simulated data was performed under systematic variation of these parameters.

RESULTS

Blood volume fitting was unreliable and affected the estimation of rate constants. Even small sample timing errors of a few seconds lead to significant deviations of the fit parameters. Data smoothing did not increase model fit precision. Accurate correction for the kinetics of blood cell uptake of FDG rather than constant scaling of the blood time-activity curve is mandatory for kinetic modeling. FDG infusion over 4 to 5 min instead of bolus injection revealed well-defined experimental input functions and allowed for longer blood sampling intervals at similar fit precisions in simulations.

CONCLUSIONS

FDG infusion over a few minutes instead of bolus injection allows for longer blood sampling intervals in kinetic modeling with the two-tissue compartment model at a similar precision of fit parameters. The fractional blood volume in the tissue of interest should be entered as a fixed value and kinetics of blood cell uptake of FDG should be included in the model. Data smoothing does not improve the results, and timing errors should be avoided by precise temporal matching of blood and tissue time-activity curves and by replacing manual with automated blood sampling.

Effect of Harderian adenectomy on the statistical analyses of mouse brain imaging using positron emission tomography

Positron emission tomography (PET) using 2-deoxy-2-[¹⁸F] fluoro-D-glucose (FDG) as a radioactive tracer is a useful technique for in vivo brain imaging. However, the anatomical and physiological features of the Harderian gland limit the use of FDG-PET imaging in the mouse brain. The gland shows strong FDG uptake, which in turn results in distorted PET images of the frontal brain region. The purpose of this study was to determine if a simple surgical procedure to remove the Harderian gland prior to PET imaging of mouse brains could reduce or eliminate FDG uptake. Measurement of FDG uptake in unilaterally adenectomized mice showed that the radioactive signal emitted from the intact Harderian gland distorts frontal brain region images. Spatial parametric measurement analysis demonstrated that the presence of the Harderian gland could prevent accurate assessment of brain PET imaging. Bilateral Harderian adenectomy efficiently eliminated unwanted radioactive signal spillover into the frontal brain region beginning on postoperative Day 10. Harderian adenectomy did not cause any post-operative complications during the experimental period. These findings demonstrate the benefits of performing a Harderian adenectomy prior to PET imaging of mouse brains.