Introduction

Placebo response—symptomatic improvement patients experience while receiving placebo during clinical trials—has recently been increasing particularly in drug trials for pediatric patients with anxiety and depressive disorders.¹ This increase in placebo response degrades the ability to detect treatment effects. Biomarkers that can identify placebo response in advance of treatment could markedly improve efficiency of clinical trials, enhance understanding of placebo response mechanisms, and potentially identify pediatric patients who may not need drug therapy. In the present study, we sought to identify biomarkers of placebo response in pediatric generalized anxiety disorder (GAD) using static and dynamic functional connectivity (FC) metrics as well as demographic and clinical features.

Methods

Twenty-five adolescents with GAD were randomized to an 8-week placebo treatment after undergoing resting-state functional and T1 structural MR imaging on a 3 T scanner with a 32-channel phased-array head coil. Three patients were excluded because of incomplete MR data, resulting in an imaging cohort of 22 patients. Demographic and clinical characteristics of the patients are shown in Table 1. Change in Pediatric Anxiety Rating Scale (PARS) score from baseline to endpoint was used to measure placebo response. Patients with greater than 35% reduction in PARS score were defined as placebo responders. Placebo responders and non-responders did not differ in age, sex, race, full-scale IQ, baseline PARS or diagnosis, nor in terms of prior treatment (Table 1).

MR data were preprocessed with the SPM12 package and DPABI toolbox using standard methods. Three regions of interest (ROI) from the amygdala-prefrontal fear circuit (amygdala, dorsal anterior cingulate cortex [dACC] and ventrolateral prefrontal cortex [VLPC]) were identified using the Brainnetome atlas.² To detect dynamic FC variability, the sliding window approach in DynamicBC toolbox was used. The sliding window length of 20 TR (40 sec) and a step of 1 TR (2 sec) produced 121 temporal windows. In each sliding window, the temporal correlation coefficients between the truncated time course of the seeds and all the other voxels were calculated. Thus, 121 correlation maps were created for each participant. Then, a Fisher’s r-to-z transformation was applied to all the correlation maps to improve the normality of the correlation distribution. The standard deviation of the z values at each voxel was used to generate dynamic FC maps. Seed‐based static FC maps were generated for each ROI using the REST package.

Whole brain voxel‐wise multiple regression analyses were performed to examine static and dynamic FC maps in relation to placebo response. The AlphaSim approach was employed to correct for multiple comparisons, with a threshold of p < 0.005 at the voxel level and p < 0.05 at the cluster level (minimum cluster size: 78 voxels). Multiple linear regression models were used to examine the relationship between demographic and clinical variables and extent of placebo response. Finally, placebo responders and non-responders were compared (2-sample t tests) with regard to static and dynamic FC parameters that were significantly associated with placebo response. p-values <0.05 were considered statistically significant.

Conclusion

Anxiety disorder is a relatively prevalent disorder that affect nearly 20% children,³ and treatment developments are needed to improve pharmacological options for this disorder. Understanding the neurobiology of placebo response is thus important. The current study is the first to examine baseline dynamic and static FC for predicting placebo response in pediatric psychiatry. Our study showed that increased placebo response was significantly associated with increased dynamic FC and decreased static FC between the amygdala, dACC, VLPFC and their functional related regions that subserve emotion control and inhibition. Moreover, there were two different brain activity patterns between placebo responders and non-responders. Specifically, increased flexibility (i.e., increased dynamic FC and decreased static FC) in placebo responders suggests that these patients may be more responsive to moment-to-moment shifts between external and internal stimuli. These findings raise the possibility that FC could serve as a useful biomarker for identifying likely placebo response, and thus be used to decrease placebo response in clinical trials to more effectively and efficiently evaluate novel treatments.

Acknowledgements

Dr. Strawn has received research support from the National Institutes of Health (NIMH/NIEHS/NICHD) as well as Allergan, Neuronetics and Otsuka. He has received material support from and provided consultation to Myriad Genetics and receives royalties from the publication of two texts (Springer) and serves as an author for UpToDate and an Associate Editor for Current Psychiatry. Dr. Strawn also receive research support from the Yung Family Foundation. Dr. Lu has receive support from the Chinese Government Scholarship. Dr. Huang has received research support from National Nature Science Foundation (Grant NO. 81671669), Science and Technology Project of Sichuan Province (Grant NO. 2017JQ0001).