Abstract
Most interventions in neuropharmacology are characterized by a striking heterogeneity of effects elicited by a drug. Hence, the effectiveness of treatment might be improved by assigning patients to tailored interventions according to the functional architecture of their brain. However, the success of personalized medicine in neuropharmacology has been limited so far. Here, we employ a novel data-driven approach, which is based on the hypothesis that similar brains will respond likewise to a dopamine challenge. To this end, we used a sample of 60 healthy participants who underwent a pharmaco-imaging study. We conducted fMRI scans at rest and during cognitive tasks after participants had received L-DOPA or placebo (randomized cross-over). Furthermore, we collected 18-fluorodopa PET scans at rest to validate the obtained results across imaging modalities. First, we characterized individual brain function using “connectomic fingerprints”, which were defined by the connectivity matrices of 136 nodes. Second, we calculated the similarity of functional connectomes across sessions and across individuals. Critically, clustering of similar connectomes identified a subset of participants, who improved their task performance after administration of L-DOPA. This cluster of benefiters was characterized by stronger functional connectivity between the striatum and the temporal lobe. In line with fMRI-based classification, benefiters were also characterized by reduced dopamine turnover in the nucleus accumbens (NAcc), which provides an independent, cross-modal validation. We conclude that the brain’s function at rest may serve as a blueprint for pharmacological effects. Hence, data-driven prediction based on similarity may help to reduce the uncertainty about prospective drug effects.
