Background
Gynecological cancer is the third most common malignancy among women in China [
1]. While improving early cancer diagnosis and accessing effective cancer treatment increase cancer patients’ five-year relative survival rate, neurocognitive dysfunctions are significant sequelae of cancer [
2]. Neurocognitive dysfunctions may affect executive function, psychomotor speed, attention and memory [
3]. This can interfere with gynecological cancer patients’ capacity to accomplish activities of daily living, as well as with social and occupational functioning, leading to lower quality of life [
2,
4,
5].
Advanced neuroimaging studies in cancer patients provide a better understanding of neurocognitive dysfunction after cancer treatment [
6]. Previous neuroimaging studies have indicated changes in brain structure and function that correlate with neurocognitive function in gynecological cancer patients [
7,
8]. While multiple neuroimaging studies have demonstrated structural and functional brain differences between cancer patients and healthy controls [
9], structural changes in the brain cannot serve as a prompt or reliable biomarker for early diagnosis of treatment-induced neurocognitive disorders [
10], as abnormalities in brain function usually appear before alterations in brain structure and clinical performance [
11]. Certainly, some studies have demonstrated that structural changes co-occur with functional network differences [
12], and the structure–function relationship is modality dependent [
13]. Other research conducted on cancer patients also found that disruptions in brain structure and/or function may parallel [
14]. Hence, detecting alterations in structural or functional brain networks might provide an earlier biomarker for neurocognitive dysfunction diagnosis [
10].
Functional magnetic resonance imaging (fMRI) studies have shown that quantitative neuroimaging techniques, in combination with neurocognitive assessment, can be useful in advancing our understanding of treatment-induced neurocognitive dysfunction in cancer patients [
7,
14,
15]. Different from task-dependent fMRI studies, resting state fMRI (Rs-fMRI) is task-independent and thus less vulnerable to confounds due to performance variance [
16]. Rs-fMRI is a noninvasive neuroimaging technique that measures spontaneous brain activity [
17]. Rs-fMRI does not require participants to engage in any cognitive activity, therefore providing unique advantages for clinical research studies [
18,
19]. In addition, rs-fMRI is sensitive enough to measure intrinsic functional networks, which reflect various cognitive states, representing the majority of energy usage in the brain [
20]. Thus, using rs-fMRI to detect brain changes pre- and post-chemotherapy is likely associated with aspects of disease and treatment pathology for cognitive dysfunction [
14]. And utilizing a network analysis of rs-fMRI data, and linked neurocognitive changes with functional brain networks in cancer patients, would be promising to address the issue of interest in the present study.
The majority of neuroimaging studies on the neurocognitive functioning of patients treated with chemotherapy for non-central nervous system cancers have been conducted on breast cancer patients [
10,
15]. Limited neuroimaging studies have been conducted on patients with gynecological cancer [
7,
8,
21‐
23]. Given the poor understanding of the impacts of cancer and its treatment on neurocognitive function and functional brain networks in gynecological cancer patients, particularly Chinese cancer patients, it is important to explore any neurocognitive changes or functional brain network alterations in this population. Therefore, this prospective longitudinal study was conducted to assess the neurocognitive function, and functional brain networks, of Chinese gynecological cancer patients pre- and post-chemotherapy. The findings could add to the literature in meaningful ways by studying a cancer type that has received limited attention in terms of the cognitive and neuroimaging effects of treatment, while adding to the small body of literature that has examined these issues in non-Caucasian patient groups.
Discussion
This is the first study to include a healthy control group with similar demographic characteristics, and a longitudinal design with repeated rs-fMRI assessment with the application of a longitudinal graph theoretical approach, to analyze brain functional networks in Chinese gynecological cancer patients. This study found that after chemotherapy treatment, gynecological cancer patients had lower neurocognitive test performance and changes in functional network measures, compared to age-matched healthy controls, which was in line with previous studies on cancer patients after chemotherapy [
10,
14]. But the mean score changes of cognitive tests in the patient group were small in this study. It may be possibly due to surgery-related cognitive impairment, as all patients at baseline received treatment of surgery. Other research suggests that cancer patients treated with local surgery yield larger cognitive impairment than patients’ own baseline [
35]. In specific, disrupted small-world properties were found in gynecological cancer patients. Functional networks with prominent small-world properties ensure higher information-processing efficiency for both locally specialized and globally integrated processing [
36]. Decreased small-worldness index among cancer patients may result in lower information processing speed, which was supported by the significant associations of lower local network efficiency with lower raw TMT-A scores.
While the findings of this study indicated that the functional brain networks of both cancer patients and healthy controls show common small-world properties (both groups’ index values > 1), the local efficiencies were significantly higher in cancer patients post-chemotherapy than in the healthy controls. As local efficiency is a measure of average local subgraphs in a network, increasing local efficiency in cancer patients may result in disrupted information processing among distant brain regions [
37], and lower network attack tolerance was associated with greater neurocognitive dysfunction in cancer patients [
14]. In addition, this study found significantly decreased global efficiency, and significantly positive correlations between decreased global efficiency and lower memory scores, in the patient group only. Study findings were consistent with previous research, which reported reduced functional brain network efficiency in response to a simulated neurodegeneration in breast cancer survivors receiving chemotherapy, compared with healthy controls [
15].
This study found that functional hubs were mostly located in the temporal regions for patients, and in the frontal and parietal regions for healthy controls, reflecting the main functions associated with these brain regions [
36]. These study findings discriminated between the functional hub networks of patients and those of healthy controls, and also identified functional hubs for patients with cognitive dysfunction as well as for patients without cognitive dysfunction. Functional hubs for patients with cognitive dysfunction included the left and right insula, middle temporal gyrus, superior temporal gyrus, left hippocampus and parahippocampal gyrus, which are essential for network resilience and regulation of information flow [
38], as functional hubs play key roles in forming bridges between different networks [
39]. Brain regions with a high node degree were identified as hubs, which would be the most vulnerable areas in local functional networks [
15]. Taken together, these findings suggest that all of these hub brain regions are key regions implicated in the pathophysiology of cognitive dysfunction; the connectome properties of these regions may to some extent predict neurocognitive functioning [
15]. Therefore, this study’s findings provide new insights into the mechanism of cognitive dysfunction in cancer patients.
Evaluating the relative importance of brain neuroimaging features and their association with neurocognitive function was essential in understanding specific brain functional network patterns involved in neurocognitive dysfunction [
15]. Rs-fMRI may be a particularly promising tool in identifying cancer patients at risk of long-term cancer-related brain injury [
14,
15]. In addition, connectome metrics derived from rs-fMRI show good test-retest reliability [
40]. Furthermore, the rs-fMRI acquisition required approximately eight minutes, making this scan a practical possibility in busy clinical settings. Thus, utilizing rs-fMRI could be a promising tool to better understand the longitudinal changes of treatment-related neurocognitive outcomes and functional network connectome properties.
The main limitation of this study is the small sample size, which may have reduced its power to detect functional differences between patients and healthy controls. This study found limited group differences achieving statistically significant differences in neurocognitive test performance, which may partially be due to limited power. Hence, in future research, there is a need to recruit larger sample sizes and use longer-term follow-up to replicate these results, and to investigate the potential reversibility of chemotherapy-induced changes [
41]. In addition, the nongeneralizable convenience sample of this study may cause potential biases that could influence the conclusions. Finally, this study only chose the AAL atlas with 90 regions (ALL-90) as a brain parcellation scheme to calculate functional connectome properties, while excluding other brain parcellation schemes, such as Harvard-Oxford Atlas, as well as randomly parceling the brain into 1024 ROIs. According to previous studies on chemotherapy-related cognitive impairment in cancer patients [
14,
15,
41], the AAL-90 parcellation is one of the most common brain parcellation schemes.