Learning to identify CNS drug action and efficacy using multistudy fMRI data.

The therapeutic effects of centrally acting pharmaceuticals can manifest gradually and unreliably in patients, making the drug discovery process slow and expensive. Biological markers providing early evidence for clinical efficacy could help prioritize development of the more promising drug candidates. A potential source of such markers is functional magnetic resonance imaging (fMRI), a noninvasive imaging technique that can complement molecular imaging. fMRI has been used to characterize how drugs cause changes in brain activity. However, variation in study protocols and analysis techniques has made it difficult to identify consistent associations between subtle modulations of brain activity and clinical efficacy. We present and validate a general protocol for functional imaging-based assessment of drug activity in the central nervous system. The protocol uses machine learning methods and data from multiple published studies to identify reliable associations between drug-related activity modulations and drug efficacy, which can then be used to assess new data. A proof-of-concept version of this approach was developed and is shown here for analgesics (pain medication), and validated with eight separate studies of analgesic compounds. Our results show that the systematic integration of multistudy data permits the generalized inferences required for drug discovery. Multistudy integrative strategies of this type could help optimize the drug discovery and validation pipeline.

Multivariate decoding of cerebral blood flow measures in a clinical model of on-going postsurgical pain.

Recent reports of multivariate machine learning (ML) techniques have highlighted their potential use to detect prognostic and diagnostic markers of pain. However, applications to date have focussed on acute experimental nociceptive stimuli rather than clinically relevant pain states. These reports have coincided with others describing the application of arterial spin labeling (ASL) to detect changes in regional cerebral blood flow (rCBF) in patients with on-going clinical pain. We combined these acquisition and analysis methodologies in a well-characterized postsurgical pain model. The principal aims were (1) to assess the classification accuracy of rCBF indices acquired prior to and following surgical intervention and (2) to optimise the amount of data required to maintain accurate classification. Twenty male volunteers, requiring bilateral, lower jaw third molar extraction (TME), underwent ASL examination prior to and following individual left and right TME, representing presurgical and postsurgical states, respectively. Six ASL time points were acquired at each exam. Each ASL image was preceded by visual analogue scale assessments of alertness and subjective pain experiences. Using all data from all sessions, an independent Gaussian Process binary classifier successfully discriminated postsurgical from presurgical states with 94.73% accuracy; over 80% accuracy could be achieved using half of the data (equivalent to 15 min scan time). This work demonstrates the concept and feasibility of time-efficient, probabilistic prediction of clinically relevant pain at the individual level. We discuss the potential of ML techniques to impact on the search for novel approaches to diagnosis, management, and treatment to complement conventional patient self-reporting.

Pharmacologic modulation of hand pain in osteoarthritis: a double-blind placebo-controlled functional magnetic resonance imaging study using naproxen.

OBJECTIVE: In an attempt to shed light on management of chronic pain conditions, there has long been a desire to complement behavioral measures of pain perception with measures of underlying brain mechanisms. Using functional magnetic resonance imaging (fMRI), we undertook this study to investigate changes in brain activity following the administration of naproxen or placebo in patients with pain related to osteoarthritis (OA) of the carpometacarpal (CMC) joint. METHODS: A placebo-controlled, double-blind, 2-period crossover study was performed in 19 individuals with painful OA of the CMC joint of the right hand. Following placebo or naproxen treatment periods, a functionally relevant task was performed, and behavioral measures of the pain experience were collected in identical fMRI examinations. Voxelwise and a priori region of interest analyses were performed to detect between-period differences in brain activity. RESULTS: Significant reductions in brain activity following treatment with naproxen, compared to placebo, were observed in brain regions commonly associated with pain perception, including the bilateral primary somatosensory cortex, thalamus, and amygdala. Significant relationships between changes in perceived pain intensity and changes in brain activity were also observed in brain regions previously associated with pain intensity. CONCLUSION: This study demonstrates the sensitivity of fMRI to detect the mechanisms underlying treatments of known efficacy. The data illustrate the enticing potential of fMRI as an adjunct to self-report for detecting early signals of efficacy of novel therapies, both pharmacologic and nonpharmacologic, in small numbers of individuals with persistent pain.

Quantifying the test-retest reliability of cerebral blood flow measurements in a clinical model of on-going post-surgical pain: A study using pseudo-continuous arterial spin labelling.

Arterial spin labelling (ASL) is increasingly being applied to study the cerebral response to pain in both experimental human models and patients with persistent pain. Despite its advantages, scanning time and reliability remain important issues in the clinical applicability of ASL. Here we present the test-retest analysis of concurrent pseudo-continuous ASL (pCASL) and visual analogue scale (VAS), in a clinical model of on-going pain following third molar extraction (TME). Using ICC performance measures, we were able to quantify the reliability of the post-surgical pain state and ΔCBF (change in CBF), both at the group and individual case level. Within-subject, the inter- and intra-session reliability of the post-surgical pain state was ranked good-to-excellent (ICC > 0.6) across both pCASL and VAS modalities. The parameter ΔCBF (change in CBF between pre- and post-surgical states) performed reliably (ICC > 0.4), provided that a single baseline condition (or the mean of more than one baseline) was used for subtraction. Between-subjects, the pCASL measurements in the post-surgical pain state and ΔCBF were both characterised as reliable (ICC > 0.4). However, the subjective VAS pain ratings demonstrated a significant contribution of pain state variability, which suggests diminished utility for interindividual comparisons. These analyses indicate that the pCASL imaging technique has considerable potential for the comparison of within- and between-subjects differences associated with pain-induced state changes and baseline differences in regional CBF. They also suggest that differences in baseline perfusion and functional lateralisation characteristics may play an important role in the overall reliability of the estimated changes in CBF. Repeated measures designs have the important advantage that they provide good reliability for comparing condition effects because all sources of variability between subjects are excluded from the experimental error. The ability to elicit reliable neural correlates of on-going pain using quantitative perfusion imaging may help support the conclusions derived from subjective self-report.

Alterations in resting-state regional cerebral blood flow demonstrate ongoing pain in osteoarthritis: An arterial spin-labeled magnetic resonance imaging study.

OBJECTIVE: Increasing evidence suggests a central nervous system (CNS) component underpinning persistent pain disease states. This study was undertaken to determine regional cerebral blood flow (rCBF) changes representing ongoing pain experienced by patients with painful osteoarthritis (OA) of the carpometacarpal (CMC) joint and to examine rCBF variability across sessions. We used pulsed continuous arterial spin labeling (pCASL), a perfusion magnetic resonance imaging (MRI) technique. METHODS: The study included 16 patients with CMC OA and 17 matched controls. Two pCASL scans and numerical rating scale (NRS) estimates of ongoing pain were acquired in each of two identical sessions. Voxelwise general linear model analyses were performed to determine rCBF differences between OA and control groups, rCBF differences between sessions within each group, and whether sessionwise rCBF differences were related to variability in perceived ongoing pain. RESULTS: In the OA group, rCBF increases representing ongoing pain were identified in the primary and secondary somatosensory, insula, and cingulate cortices; thalamus; amygdala; hippocampus; and dorsal midbrain/pontine tegmentum, including the periaqueductal gray/nucleus cuneiformis. Sessionwise rCBF differences in the OA group in the postcentral, rostral/subgenual cingulate, mid/anterior insula, prefrontal, and premotor cortices were related to changes in perceived ongoing pain. No significant sessionwise rCBF differences were observed in controls. CONCLUSION: This is the first quantitative endogenous perfusion MRI study of the cerebral representation of ongoing, persistent pain due to OA. Observed rCBF changes potentially indicate dysregulated CNS appraisal and modulation of pain, most likely the maladaptive neuroplastic sequelae of living with painful OA. Understanding the neural basis of ongoing pain is likely to be important in developing novel treatment strategies.

Accuracy and reliability of microbubble ultrasound measurements for the non-invasive assessment of hepatic fibrosis in chronic hepatitis C.

AIM: Contrast-enhanced ultrasound can be used to assess liver disease severity non-invasively by observing intra- and extrahepatic hemodynamic changes. Transit times are calculated to include intra- and extrahepatic components (hepatic vein transit time, HVTT) or the intrahepatic component (hepatic transit time, HTT), but these have not been compared directly. We aimed to compare diagnostic accuracy of HVTT and HTT in gauging the severity of chronic hepatitis C (CHC) and to assess inter- and intra-observer reliability. METHODS: Recorded ultrasound scans performed on 75 patients with biopsy-staged CHC, using the microbubble contrast agent Sonovue were analyzed. RESULTS: Diagnostic accuracy of HTT and HVTT for diagnosis of cirrhosis was 0.78 and 0.71 (P = 0.24). Diagnostic accuracy of HTT and HVTT for diagnosis of fibrosis stage >2 was 0.76 and 0.72 (P = 0.23). Negative predictive value for cirrhosis using this cut-off was high for both techniques (HVTT, 88%; HTT, 92%), suggesting utility for exclusion of cirrhosis. Inter-observer reliability for HTT and HVTT were 0.92 and 0.94, respectively. Intra-observer reliability for HTT and HVTT were 0.98 and 0.99. CONCLUSION: In this cohort, reliability exceeded 90% while diagnostic accuracy was in keeping with previous studies of microbubble transit time analysis. Despite higher numerical diagnostic accuracy for HTT, no significant difference was demonstrated between the techniques, suggesting that both methods can be used reliably.

The response to rapid infusion of fentanyl in the human brain measured using pulsed arterial spin labelling.

OBJECTIVE: We evaluated the sensitivity of pulsed Arterial Spin Labelling (pASL) for the detection of changes in regional cerebral blood perfusion (CBP) during and after intra-venous (i.v.) infusion of an opioid agonist (fentanyl) and an opioid antagonist (naloxone). MATERIALS AND METHODS: Twenty-three subjects were scanned four times, receiving i.v. infusion of fentanyl, naloxone, placebo and a second fentanyl administration, in four separate scanning sessions in randomised order. End-tidal CO(2), respiration rate and heart rate were recorded continuously throughout each scan. pASL time series were collected using single shot EPI for 15 min (including 5 min of baseline prior to infusion). RESULTS: Significant increases in CBP were detected during and after administration of fentanyl, (when compared to placebo and naloxone), in most areas of high concentration of mu-opioid receptors (thalamus, lingual gyrus, para-hippocampal gyrus, and insula); near-significant increases were also observed in the insula. No increases in perfusion were observed during or after naloxone infusion. No correlation was found between regional rCBF changes and end-tidal CO(2), respiration rate or heart rate. Good reliability was found between the first and second fentanyl sessions but the regions of high reliability did not overlap completely with those of highest perfusion change. CONCLUSION: pASL is a suitable method for examining rapid, dynamic effects of opioid administration on brain physiology.

Applications of multi-nuclear magnetic resonance spectroscopy at 7T.

AIM: To discuss the advantages of ultra-high field (7T) for (1)H and (13)C magnetic resonance spectroscopy (MRS) studies of metabolism. METHODS: Measurements of brain metabolites were made at both 3 and 7T using (1)H MRS. Measurements of glycogen and lipids in muscle were measured using (13)C and (1)H MRS respectively. RESULTS: In the brain, increased signal-to-noise ratio (SNR) and dispersion allows spectral separation of the amino-acids glutamate, glutamine and γ-aminobutyric acid (GABA), without the need for sophisticated editing sequences. Improved quantification of these metabolites is demonstrated at 7T relative to 3T. SNR was 36% higher, and measurement repeatability (% coefficients of variation) was 4%, 10% and 10% at 7T, vs 8%, 29% and 21% at 3T for glutamate, glutamine and GABA respectively. Measurements at 7T were used to compare metabolite levels in the anterior cingulate cortex (ACC) and insula. Creatine and glutamate levels were found to be significantly higher in the insula compared to the ACC (P < 0.05). In muscle, the increased SNR and spectral resolution at 7T enables interleaved studies of glycogen ((13)C) and intra-myocellular lipid (IMCL) and extra-myocellular lipid (EMCL) ((1)H) following exercise and re-feeding. Glycogen levels were significantly decreased following exercise (-28% at 50% VO(2) max; -58% at 75% VO(2) max). Interestingly, levels of glycogen in the hamstrings followed those in the quadriceps, despite reduce exercise loading. No changes in IMCL and EMCL were found in the study. CONCLUSION: The demonstrated improvements in brain and muscle MRS measurements at 7T will increase the potential for use in investigating human metabolism and changes due to pathologies.

Beyond patient reported pain: perfusion magnetic resonance imaging demonstrates reproducible cerebral representation of ongoing post-surgical pain.

Development of treatments for acute and chronic pain conditions remains a challenge, with an unmet need for improved sensitivity and reproducibility in measuring pain in patients. Here we used pulsed-continuous arterial spin-labelling [pCASL], a relatively novel perfusion magnetic-resonance imaging technique, in conjunction with a commonly-used post-surgical model, to measure changes in regional cerebral blood flow [rCBF] associated with the experience of being in ongoing pain. We demonstrate repeatable, reproducible assessment of ongoing pain that is independent of patient self-report. In a cross-over trial design, 16 participants requiring bilateral removal of lower-jaw third molars underwent pain-free pre-surgical pCASL scans. Following extraction of either left or right tooth, repeat scans were acquired during post-operative ongoing pain. When pain-free following surgical recovery, the pre/post-surgical scanning procedure was repeated for the remaining tooth. Voxelwise statistical comparison of pre and post-surgical scans was performed to reveal rCBF changes representing ongoing pain. In addition, rCBF values in predefined pain and control brain regions were obtained. rCBF increases (5-10%) representing post-surgical ongoing pain were identified bilaterally in a network including primary and secondary somatosensory, insula and cingulate cortices, thalamus, amygdala, hippocampus, midbrain and brainstem (including trigeminal ganglion and principal-sensory nucleus), but not in a control region in visual cortex. rCBF changes were reproducible, with no rCBF differences identified across scans within-session or between post-surgical pain sessions. This is the first report of the cerebral representation of ongoing post-surgical pain without the need for exogenous tracers. Regions of rCBF increases are plausibly associated with pain and the technique is reproducible, providing an attractive proposition for testing interventions for on-going pain that do not rely solely on patient self-report. Our findings have the potential to improve our understanding of the cerebral representation of persistent painful conditions, leading to improved identification of specific patient sub-types and implementation of mechanism-based treatments.

A comparison of single-voxel clinical in vivo hepatic 31P MR spectra acquired at 1.5 and 3.0 Tesla in health and diseased states.

With the increasing availability of human MR scanners at various field strengths, the optimal field strength for in vivo clinical MR studies of the liver has become a focus of attention. Comparison between results at 3.0 and 1.5 T is of particular clinical interest, especially for multicentre studies. For MRS studies, higher field strengths should be advantageous, because improved sensitivity and chemical shift dispersion are expected. We report a comparison between single-voxel hepatic proton-decoupled (31)P MRS performed at 1.5 and 3.0 T in the same subjects using similar methodologies. Twelve healthy volunteers and 15 patients with chronic liver disease were studied. Improved spectral resolution was achieved using proton decoupling, and there was an improvement (21%) in the signal-to-noise ratio (SNR) of the phosphomonoester (PME) resonance at 3.0 T relative to 1.5 T. There was no significant change in nuclear Overhauser effects for PME or phosphodiesters (PDEs) between the two field strengths. The T(1) value of PDE was significantly longer at 3 T, although there was no significant change in the T(1) value of PME. There was no significant difference in the mean PME/PDE ratios for either the control or patient groups at both 1.5 and 3.0 T, but there was a small positive mean difference in PME/PDE at 3.0 T on pairwise testing between field strengths (+ 0.05, p < 0.01). There were significant correlations between PME/PDE values at the two magnetic field strengths (r = 0.806, p < 0.001). The underlying broad resonance was reduced at 3.0 T relative to 1.5 T, related to line broadening of the phospholipid bilayer signal. In conclusion, there was an improvement in hepatic (31)P MR signal quality at 3.0 T relative to 1.5 T. Broadly similar hepatic (31)P MR parameters were obtained at 1.5 and 3.0 T. The modest difference noted in the PME/PDE ratio between field strengths for patients with chronic liver disease should inform multicentre study design involving these field strengths.

Carotid artery hemodynamics: observing patient-specific changes with amlodipine and lisinopril by using MR imaging computation fluid dynamics.

PURPOSE: To assess whether using magnetic resonance (MR) imaging combined with computational fluid dynamics (CFD) could reveal changes in common carotid artery (CCA) flow and wall shear stress (WSS) that might contribute to differences in CCA remodeling between amlodipine, a calcium channel blocker, and lisinopril, an angiotensin-converting enzyme inhibitor, despite similar reductions in blood pressure (BP). MATERIALS AND METHODS: Institutional review board approval was obtained, and participants gave informed consent. Nine subjects with hypertension were recruited into a double-blind placebo-controlled randomized three-way crossover study to compare the hemodynamic effects of 7 days of treatment with placebo, amlodipine, or lisinopril. After each treatment period, patients underwent CCA ultrasonography, BP measurement, and MR imaging with CFD. Analyses were performed by using repeated-measures analysis of variance, followed by the Tukey test or the Wilcoxon matched-pairs test. RESULTS: Amlodipine and lisinopril lowered BP similarly, but CCA flow rate was significantly higher (P < .01) and distal vascular resistance was lower (P = .016) after amlodipine treatment than after lisinopril treatment. WSS on the inner wall of the CCA was significantly lower after lisinopril treatment than after amlodipine treatment (P = .03). The change in WSS in the CCA correlated with the change in vascular resistance (r = -0.85, P < .001). CONCLUSION: Amlodipine causes increased blood flow and increased time-averaged WSS in the CCA compared with lisinopril, despite similar reductions in BP. Differences in the subacute hemodynamic effects of amlodipine and lisinopril could contribute to the differences in CCA remodeling seen in long-term studies. SUPPLEMENTAL MATERIAL: http://radiology.rsna.org/lookup/suppl/doi:10.1148/radiol.10100788/-/DC1.

Information content and quality of MR thumb images.

This study investigates the relation between objective and subjective parameters which relate to the quality of magnetic-resonance images, with a view to developing procedures for minimizing imaging time. The investigation uses high-resolution images of the thumb as example. Detection thresholds for an artificial lesion and ratings of image quality were obtained for a variety of images, with several experienced observers. In addition, the information content was calculated for each image, using the method developed by Fuderer (1988 IEEE Tralns. Med. Imaging 7 368-80). Results suggest that information content can be used as a predictor of either detection threshold or quality rating, with a critical information content beyond which there is no significant improvement in either of these aspects of image quality. Since it is possible to estimate the effect of imaging variables on information content, it is possible to predict the effect of time-saving imaging strategies on image quality. A procedure is suggested for determining the combination of imaging variables which gives the shortest possible imaging time whilst retaining image quality.