Annual smoking-attributable deaths was about 6 million in the world including one million in China (World Health Organization, 2013). “China Report on the Health Hazards of Smoking” in 2012 indicated China had more than 350 million smokers, of which 14 millions were young smokers. The prevalence of smoking during adolescence was higher than other age stages. People during the period from adolescence to young adulthood are undergoing a series of continuous and important physiological development. Therefore, young people are relative sensitive to outside influences, especially the effects of nicotine exposure (Gulley and Juraska, 2013; Lydon et al., 2014).

During the resting state, the brain may be conceived as a complex functional network constituted by correlated spontaneous fluctuations occur within brain regions (Seeley et al., 2009). Electroencephalography (EEG) had been employed to measure ongoing oscillation of neural electrical activity in the absence of an overt task or stimulation (Delbeuck et al., 2003; Stam et al., 2007). By using spectral analysis of neural oscillations in EEG, previous resting state studies had found decreased theta and delta power, increased beta power and inconsistent results in alpha band.in smokers (Rass et al., 2015). Graph theoretical network analyses (GTA) of EEG signal during the resting state was used to investigate the integrity and efficiency of global brain networks in brain diseases, by assessing the topology characteristics of global brain networks (Delbeuck et al., 2003; Lee et al., 2006; Leistedt et al., 2009; Fraga González et al., 2016). The findings improved our understanding of neural mechanisms of brain diseases from whole brain network. However, few study used EEG to investigate the changes of global brain networks during the resting state in young smokers.

Spectral analysis provided us a way to detect the changes of spectral power in different neural oscillations, which were related with the cortical arousal or activation (Rass et al., 2015). Compared with traditional EEG methods, the phase lag index (PLI) assessed the consistency of phase lags between time series, which may avoid the effect of spurious interactions caused by common reference effects or volume conduction (Stam et al., 2007; Yuan et al., 2016). Moreover, the Minimum Spanning Tree (MST) analysis can be used to derive the connectivity matrix of functional network topology, which contains the strongest connections to avoid the problem of arbitrary thresholding of connectivity values for network reconstruction and preserve important information about network organization (Boersma et al., 2013; Tewarie et al., 2015). Therefore, in the current study, we constructed global network of EEG in different spectrum bands to investigate the changes of functional network topology during the resting state in young smokers. We used PLI to construct the matrix of functional connectivity and employed MST to reconstruct whole-brain functional connectivity networks of young smokers in different frequency bands (Brody et al., 2007; van Dellen et al., 2014, 2015; Tewarie et al., 2015; Fraga González et al., 2016). Main four EEG bands (delta, theta, alpha, and beta) with global brain recording sites were analyzed to examine the differences of global brain network topology between young smokers and nonsmokers. It is hoped that our study could improve the understanding of neural mechanisms in smoking.

Complex brain networks may play a key role in the integrated processing of the brain function, which may be influenced by pathological processing. In present study, we analyzed EEG signals during the resting state to examine the topological characteristics of brain networks in young smokers by using PLI and MST methods. The major problem of EEG was volume conduction, which may cause overestimation of functional connectivity. The PLI measured the consistency of phase lags between EEG time series and provide more accurate results of functional connectivity (Stam et al., 2007; van Dellen et al., 2015; Yuan et al., 2016). Moreover, the MST method assessed the strongest connections in the network without forming loops, which solved the problem of arbitrary thresholding of connectivity values and preserved important information about network organization (Tewarie et al., 2015). Compared with nonsmokers, young smokers showed decreased leaf fraction, kappa value and increased diameter, eccentricity value in alpha band.

Leaf fraction indicates the extent of central organization in global network. High leaf fraction suggests that the communication of network is largely dependent on hub nodes. Kappa value indicates the broadness of the degree distribution (degree divergence). Diameter is the largest distance between any two nodes of the tree, which is related to the efficiency of global brain network organization. Information may be efficiently processed between remote brain regions with low diameter. In addition, eccentricity indicates longest distance between a reference node and any other node, which reflects the efficient information processing from the least central node (van Dellen et al., 2015; Fraga González et al., 2016). In present study, decreased leaf fraction, kappa value were found in young smoker, which were similar with previous EEG studies brain neural oscillations properties in smokers during the resting state (Rass et al., 2015). Decreased alpha band power in smokers might be associated with long-term consequences of nicotine use (Rass et al., 2015; Wilbanks et al., 2016). Despite of the significant advantages, the PLI analyses failed to reveal differences of the connectivity strength between nodes of the brain regions. The MST method avoided potential bias as thresholding value (Stam, 2014), which constructed of brain connectivity network (van Diessen et al., 2014). The MST analysis had been successfully applied to EEG data from several diseases (Delbeuck et al., 2003; van Dellen et al., 2014; Fraga González et al., 2016). In the current report, we found that smokers were associated with altered global brain networks. We also revealed the abnormal connectivity strength in the alpha band in young smokers by using the MST analyses. The optimal tree was configured by both short distances and no overload of the central tree nodes (Stam, 2014). In our study, the MST analysis presented us a disordered network of smokers as marked by decreases in leaf number indicating low network integration compare with nonsmokers. These results suggested a more path-like configuration in young smoker and a more star-like topology in nonsmoker in the tree topologies. The MST leaf fraction was lower for trees derived from regular networks and increased as these networks became more random and increased distances (Tewarie et al., 2015). Increased diameter value emerge to less communication between nodes of the network in smokers, which suggested the brain networks became more disorganized and decreased global efficiency in smokers. The eccentricity and kappa descript the configuration features of tree. In present study, increased eccentricity may reveal increased longest distance between a reference node and any other node in young male smokers, which reflects decreased efficient information processing from the least central node. Decreased kappa value was related to decreased broadness of the degree distribution, which may suggest slow synchronization of the tree in young male smokers (Stam and van Straaten, 2012).

Our results indicated less integrated network configuration in young smokers, which confirmed the decrease of global efficiency in alpha band in young smokers compared with nonsmokers. These findings may be associated with cholinergic deficits and cognitive impairment in young smokers. Positive correlation was found between leaf fraction and onset times in young smoker, which may suggest initial smoking was associated with the extent of central organization in global network. Young male smokers were associated with changes in brain network topology, including further decreased leaf number and increased eccentricity in the brain.

The study is limited in its cross-sectional design, and it is not possible to infer the causality of the relationships between alpha band and severity of smoking. The relationships might be bi-directional and related to some latent variables. In addition, our participants were all male. Previous studies revealed clear gender effect in smoking effects (Kühn et al., 2012; Franklin et al., 2014). Another limitation of the study is the relatively small sample size, so the results might not be generalize able to wider populations. Further longitudinal studies with larger sample size would be required to improve the understanding of the EEG findings in the current study.

The abnormal topological characteristics of functional brain networks in alpha band were found in young smokers during the resting state. The findings may provide novel insights into the neural mechanism in smoking. We hoped our study could improve the understanding of smoking.

SS, DY, KY, and XL conceived and designed the experiments; KY, SS, JC, YC, XZ, and YG performed the experiments; DY, KY, JC, YC, and TX analyzed the data; SS, DY, and KY wrote the article; YL and YB provided critical revision of the manuscript for important intellectual content. All authors critically reviewed content and approved final version for publication.