Lung cancer ranks first in the global cancer morbidity and mortality ranking, accounting for 11.6% and 18.4% of all incidences of cancer in the population, respectively. Lung cancer can be divided into small cell lung cancer (SCLC) and non-small cell lung cancer (NSCLC), while NSCLC accounts for more than 85% of the total lung cancer cases (Bray et al. 2018). Chemotherapy, radiotherapy, surgery and targeted therapy are commonly used in the clinical treatment of lung cancer patients (Bray et al. 2018). In recent years, a number of neuropsychological studies have shown that lung cancer and chemotherapy have certain negative influences on the executive function and cognitive memory function of patients’ brains (Grosshans et al. 2008; Simo et al. 2013; Simo et al. 2018; Simo et al. 2015; Simo et al. 2016). In previous studies, short transient cognitive impairment in patients with NSCLC after chemotherapy was found (Kaasa, Mastekaasa, and Naess 1988; Simo et al. 2016), and nearly 60–90% of patients with SCLC had cognitive impairment after chemotherapy for 1 to 5 months. Neuroimaging studies have also demonstrated chemotherapy-related brain structural and functional changes in lung cancer patients (Simo et al. 2013; Simo et al. 2018; Simo et al. 2015; Simo et al. 2016). However, whether NSCLC and chemotherapy have a broad effect on brain networks in patients with NSCLC remains unclear.
Previous studies have shown that lung cancer tumors themselves can lead patients to develop paraneoplastic neurological syndrome (PNS) (Graus and Dalmau 2012; Honnorat and Antoine 2007; Leypoldt and Wandinger 2014; Voltz 2002), and cognitive impairment known as “chemobrain” could also be caused by chemotherapy in cancer patients (Ahles and Saykin 2007; Simo et al. 2013; Zeng et al. 2020). First, PNS causes the nervous system distancing effect, which causes symptoms of damage in some parts of the central nervous system, leading to cognitive dysfunction in patients (Leypoldt and Wandinger 2014). The mechanism may be that malignant tumor cells change inflammatory mediators, such as cytokines, chemokines, platelets and neutrophils, in the body environment through the inflammatory response of the tumor microenvironment during the growth process and promote the metabolites of tumor cells to cross the blood–brain barrier and affect the brain (Graus and Dalmau 2012; Leypoldt and Wandinger 2014). Second, drug molecules across the blood–brain barrier react to neurotoxicity in the brain during the process of chemotherapy, and neurons with oxidative damage, cell factor disorders and nerve repair and/or plasticity-related genes of individual variation and other factors may lead to a brain structure in patients with a wide range of morphological changes, functional brain damage, and cognitive impairment (Ahles and Saykin 2007; Hosseini., Koovakkattu., and Kesler. 2012; Janelsins et al. 2011; Wefel, Witgert, and Meyers 2008). In addition, chemotherapeutic drug molecules may induce ischemic lesions, hinder the formation of cerebrovascular blood and reduce cerebrovascular blood flow, thus causing local lesion and functional damage to the cerebral cortex of patients (Simo et al. 2013). Third, recent neuroimaging studies have also found changes in the brain structure networks of lung cancer patients. For example, Liu et al. used diffusion-tensor imaging (DTI) of structural connectivity to analyze the connection network using the graph theory method and found that the topological characteristics of the structural network in the brain of patients with NSCLC before chemotherapy were impaired, the clustering coefficient of the left hippocampus was significantly reduced, and cognitive function was reduced (Liu et al. 2020). Using magnetic resonance imaging (MRI), Simo et al. found that lung cancer patients had speech memory deficits and extensive white matter damage before chemotherapy, while reduced gray matter density and impaired white matter integrity showed impaired visuospatial and verbal fluency after chemotherapy (Simo et al. 2015). Using functional magnetic resonance imaging (fMRI), Simo et al. also found that cognitive deficits induced by lung cancer and chemotherapy resulted in abnormal functional connections related to the default network, left and right anterior temporal lobe network, and cerebellar network and decreased connectivity (Simo et al. 2018). In brief, lung cancer patients’ specific brain networks, which are involved in various functions, such as executive function and cognitive memory, seem to be most vulnerable.
To date, most studies have focused on exploring the changes in lung cancer patients’ brain networks through DTI and fMRI connections (Bromis et al. 2017; Liu et al. 2020), while few methods have been used to construct networks based on the morphological similarities of the gray matter cortex. Cortical morphological data contain a large amount of brain structural connectivity information, while brain structure and brain function interact with each other, and the functional state of the brain may come from the basis of similar brain region structure and morphology (Rubinov and Sporns 2010). The morphological similarity network is based on the structural similarity between each pair of nodes. The nodes represent the cerebral cortex area and are considered to be connected when they are on the cortical thickness or volume covariant, or when they are shown in a single subject’s structural similarity(Tijms et al. 2012; Tijms et al. 2013). Morphological similarity may be a measure of the degree of synchronous development and maturity of brain regions in subjects, reflecting the synergistic change pattern of brain regions’ structures; the more similar the structures are, the stronger the connectivity of brain regions is (Alexander-Bloch, Giedd, and Bullmore 2013; He et al. 2009). At present, most studies define the nodes in the brain structural network by the brain regions divided by prior templates, but the brain regions as nodes cannot include more detailed morphological features. Furthermore, the structural network is calculated based on the population level, so the results of networks are at the group level, while the structural network changes for a single individual cannot be measured (Liu et al. 2020; Simo et al. 2018; Simo et al. 2015). Therefore, in our work, individual morphological similarity networks of lung cancer patients were constructed to explore the effects of chemotherapy on patients.
In this study, we first hypothesized that there may be chemotherapy-induced brain structural network changes in patients with NSCLC receiving chemotherapy. The morphological similarity networks (Tijms et al. 2012) of the gray matter were constructed from the NSCLC patients receiving/not receiving chemotherapy and healthy control groups. Then, graph-theoretical properties based on the morphological similarity networks were calculated and compared among the above 3 groups. In addition, relations between those network characteristics and clinical parameters, including granulocytes and thrombocytes, were also investigated.