Functional Connectivity in the Pharmacologically Activated Brain: Resolving Networks of Correlated Responses to d-Amphetamine Adam J. Schwarz, * Alessandro Gozzi, Torsten Reese, and Angelo Bifone We investigated the functional connectivity structure underly- ing the widespread relative cerebral blood volume (rCBV) re- sponse to d-amphetamine in the rat brain by systematically analyzing the intersubject correlations between the response amplitudes in 48 atlas-defined brain structures. A cluster anal- ysis resolved three distinct networks of brain regions that ex- hibited closely coupled responses: one corresponding to pri- mary dopamine projections from the midbrain to the striatum, a second consisting predominantly of forebrain cortical and basal ganglia regions that share a widespread correlation pattern resembling the univariate group response, and a third including structures in the periventricular dopamine system. These re- sults suggest that different functional networks underlie the brain’s response to d-amphetamine. This approach may pro- vide important new insights regarding the central systems that underlie pharmacological action. Magn Reson Med 57: 704 –713, 2007. © 2007 Wiley-Liss, Inc. Key words: phMRI; cluster analysis; functional connectivity; amphetamine; rat In recent years functional MRI (fMRI) methods have been increasingly used to examine the brain’s response to phar- macological action by individual drugs of abuse or com- pounds targeted at specific receptor sites (1–3). One aspect of these endeavors is to study the direct central effects of pharmacological administration on the hemodynamic re- sponse, possibly modulated by disease state or pretreat- ment. This pharmacological MRI (phMRI) approach, which quantifies the response to an acute drug challenge, has been employed in both preclinical and human studies to obtain an in vivo, system-level view of drug effects on the brain. However, while such experiments provide a measure of the brain’s response to pharmacological action, the ob- served signal changes can potentially derive from a num- ber of sources, including effects at the primary site of action of the drug and downstream activation in other parts of the brain. Standard massively univariate analyses of phMRI signal changes (independent group comparisons at each image voxel) do not distinguish between these effects, and even drugs that are selective for a particular target can give rise to a widespread central response. A complementary approach is to explicitly examine corre- lated signal changes between different brain regions, since the existence of correlated changes may be interpreted as a sign of underlying functional connectivity (4). This ap- proach has been applied most frequently to signal changes during cognitive tasks or “resting-state” brain activity, and has recently been applied to study functional connectivity in the acute central response to a drug challenge per se (5–7). The identification of interactions between brain struc- tures in response to drug administration has the potential to tell us much about the central substrates underlying drug action (6,7). In the present article we report a system- atic profile of the functional connectivity structure under- lying the central relative cerebral blood volume (rCBV) changes following d-amphetamine challenge in the rat. d-Amphetamine has been widely used in phMRI experi- ments as a systemic stimulus of the dopaminergic system. It gives rise to a rather widespread response involving both cortical and subcortical regions when administered acutely (8 –10). We were interested in disentangling the functional networks of brain regions recruited in response to d-amphetamine by examining how they are correlated in this response. To allow a tractable analysis of the rela- tionships between the responses in different brain regions, we parcellated the brain into 48 atlas-based bilateral struc- tures. A correlation analysis on the intersubject response vectors in these regions revealed a rich structure underly- ing the widespread signal changes detected via a standard group comparison. To clarify this complex connectivity pattern, we performed a cluster analysis to identify groups of structures that were closely coupled in their response. We show that the overall correlation structure can be re- solved into three subnetworks of brain regions that are closely coupled in their rCBV response to d-amphetamine. MATERIALS AND METHODS Animal Preparation All of the experiments were carried out in accordance with Italian regulations governing animal welfare and protec- tion. The protocols were also reviewed and consented to by a local animal-care committee in accordance with the guidelines of the Principles of Laboratory Animal Care (NIH publication 86-23, revised 1985). Twenty-four male Sprague-Dawley rats were surgically prepared and moni- tored as previously described (8,11) and imaged under 0.8% halothane maintenance anesthesia, neuromuscular blockade (d-tubocurarine), and artificial ventilation. The ventilation parameters were adjusted for each animal to ensure that its blood gas values remained within physio- logical range. Blood pressure and temperature were mon- itored continually throughout the experiment. Department of Neuroimaging, Psychiatry Center of Excellence in Drug Dis- covery, GlaxoSmithKline, Verona, Italy. *Correspondence to: Adam J. Schwarz, Department of Neuroimaging, Section of Biology, Psychiatry Center of Excellence in Drug Discovery, GlaxoSmithKline, Via Fleming 4, 37135 Verona, Italy. E-mail: adam.j.schwarz@gsk.com Received 6 July 2006; revised 29 November 2006; accepted 3 December 2006. DOI 10.1002/mrm.21179 Published online in Wiley InterScience (www.interscience.wiley.com). Magnetic Resonance in Medicine 57:704 –713 (2007) © 2007 Wiley-Liss, Inc. 704