The present study provides an initial window into the neural mechanisms underlying CRCI of BC patients. Using a sample comprised of NAC-treated BC patients and two HC cohorts at the same time, we prospectively investigated changes associated with NAC in vivo in indices of the functional brain connectome that may be related to cognitive dysfunction, excluding population baseline differences and temporal interference. Consistent with our hypothesis, early NAC caused acute damage to the brain functional connectome in BC patients, which manifested as disruption of the brain network after the first cycle of NAC. Importantly, shifts in CRCI-related topological properties were correlated with changes in clinical scales, more directly supporting the idea that brain functional connectome properties can reflect key pieces of the neural mechanisms by which NAC impacts symptoms.
At the global level, compared with tp1, BC patients who experienced or were experiencing NAC exhibited a more randomized brain network both at tp2 and tp3, in which the brain network transforms from a small-world pattern to a relatively random network pattern. The occurrence of this pattern in our results is consistent with our first hypothesis and demonstrates that NAC could alter the brain functional connectome in BC patients. Some neuroimaging studies involving application of also applying connectome-based models on fMRI and structural MRI found a disrupted brain network in BC patients after chemotherapy [30, 36, 37]. Such a pattern of brain networks has been also improved in some neuropsychiatric disorders, such as posttraumatic stress disorder [21] and essential tremor [24], reflecting a less optimal topological organization in brain networks. In our study, the disrupted network pattern was characterized by an increased Eglob (efficiency of exchanging information at the global level [38]) and decreased Lp (the average distance from one node to any other node in the network, expressed as the number of links that must be traveled [25]), which could lead to information transfer disorder and constrain long-distance functional integration as reflected by the abnormal information processing speed and disrupted executive capacity of BC participants after chemotherapy [39].
At the nodal level, both tp2 and tp3 showed significant differences compared with tp1 that were related to the frontal-limbic system (right superior frontal gyrus, right inferior temporal gyrus, left frontal gyrus, left amygdala) and cerebellar regions. The frontal-limbic system is believed to support emotion processing and regulation, reward processing, and cognitive control [40, 41]. Cerebellar regions are also increasingly recognized as important in higher cognitive functions [42]. Our observations in the frontal-limbic system suggested that alterations in this circuitry may play an important role in the neurobiology of cognitive dysfunction and mood dysregulation via impact on cognition and emotional reactivity [43]. As a key brain region in this circuit, the amygdala can not only regulate emotional and cognitive functions through the interaction of the limbic system or other brain regions [44], but also balance the brain microenvironment, which is dominated by inflammatory response, thus affecting the peripheral clinical symptoms of individuals [45]. Some studies have shown that BC and its treatment may activate microglia, which remain preactivated in brain regions, mainly the amygdala, even after completion of treatment and resolution of peripheral inflammation [46]. As has been shown in developmental [47] and aging models [48], such priming may increase neural sensitivity to peripheral inflammation with consequences for neural function, cognition, and behavior. This finding was consistent with the positive correlation between the change in nodal parameters of the amygdala and the change in PCA obtained in this study, which indicates that early NAC will bring an acute inflammatory response at the physiological level [49], and this response may be reflected as cognitive impairment at the individual level.
The frontal gyrus in the frontal-limbic system is known to be involved in cognitive processing [50]. McDonald et al. [51] reviewed the currently published articles in the changes of brain structure in BC patients after chemotherapy, and the results revealed that the frontal gyrus showed a significant change after chemotherapy that was related to a failure to regulate cognition-related processes. Moreover, functional neuroimaging studies have also shown abnormalities after chemotherapy, manifested as a decrease in gray matter density in the frontal gyrus, which is consistent with our findings (decreased centrality of the frontal gyrus) [52, 53]. In addition, it has been suggested that there is a cognitive neurobiological model demonstrating that functional deficits in the frontal gyrus may impair top-down control over the limbic areas [54]. However, the changes in the frontal gyrus only appeared after the first cycle of NAC. This was consistent with longitudinal studies suggesting that the frontal gyrus gradually recovers over time after early chemotherapy-related acute injury [16, 39]. Taken together, alternations in the frontal-limbic system may be associated with acute impairment of cognitive regulation in BC patients after early chemotherapy.
Some limitations must also be considered. First, the brain parcellation template we selected for constructing brain networks may affect the network analysis results. As different templates may lead to different estimates of graph theory parameters, this factor needs to be considered [55]. Second, the sample size of participants in our study was relatively small. Third, our study lacked a BC control group without NAC treatment. In addition, the HC groups in our study consisted of two cohorts that were assessed at two time points (tp1 vs. tp2 and tp1 vs. tp3) due to some people refusing follow-up. Since there was no difference within the HC group and no difference between HC and BC at baseline, we basically ruled out the aging effects on topological properties, and indicated that the tumor itself had no significant effect on topological properties at baseline. In the future, we will try to include and expand healthy controls with data acquisition time intervals identical to those of BC patients who received NAC. Fourth, in this study, the FACT-Cog, the clinical scale we used for the assessment of cognitive function, is not a direct objective measure. We will pay attention to and correct this problem in future clinical-scale collection processes.