The pathways through which chemotherapeutic agents induce cognitive dysfunction are related to their differential effects on the blood–brain barrier, which in turn leads to varied neuroimaging manifestations. Studies indicate that patients treated with anthracycline-containing regimens show significantly elevated serum levels of pro-inflammatory cytokines such as TNF-α and IL-6. The levels of these inflammatory markers are significantly associated with increased subjective cognitive complaints and declines in objective neuropsychological test performance (Cheung et al., 2015; Janelsins et al., 2017).

Platinum-based agents exhibit different properties. The choroid plexus, a key component of the blood-cerebrospinal fluid barrier and the primary site of CSF production, may be involved. Platinum drugs are known primary etiological agents for peripheral neuropathy, with mechanisms potentially involving direct damage to dorsal root ganglia. This damage suggests that platinum drugs may more directly affect the central nervous system microenvironment, subsequently impacting adjacent brain regions such as the hippocampus. Research by Simó et al. demonstrated decreased gray matter density in multiple brain regions, including temporolimbic structures like the hippocampus and parahippocampal gyrus crucial for cognitive function, in lung cancer patients receiving platinum-based chemotherapy (Cheung et al., 2015). This finding directly links drug characteristics to observable structural changes in the human brain.

Studies show that a higher number of courses and doses of methotrexate treatment leads to a decreased prevalence of leukoencephalopathy in acute lymphoblastic leukemia (ALL) survivors after one and a half years (Reddick et al., 2005). CSF analysis suggests that long-term elevation of axonal injury markers like myelin basic protein (MBP) in survivors provides molecular-level support for persistent white matter damage (Cheung et al., 2018). In neuroimaging, this microstructural damage manifests as alterations in diffusion tensor imaging (DTI) parameters. Zou et al. (2017) observed decreased fractional anisotropy (FA) and increased radial diffusivity (RD) in the white matter of ALL survivors who received chemotherapy. Notably, these alterations in white matter integrity and associated cognitive difficulties remain detectable years after treatment cessation, indicating potentially long-term effects. However, studies by Baratto and Shai et al. suggest that high-dose methotrexate in hematologic malignancies like lymphoma may induce more diffuse white matter damage, appearing as reduced glucose metabolism and blood flow in specific brain regions on imaging. They demonstrated that the mean standardized uptake value (SUVmean) and mean cerebral blood flow (CBFmean) in the prefrontal cortex and cingulate gyrus could quantitatively detect executive cognitive function (Baratto et al., 2024; Shrot et al., 2019). Specific results require further analysis.

Prostate cancer has surpassed lung cancer as the most common cancer in men globally and a leading cause of male cancer death (Bray et al., 2024). Research indicates that cognitive dysfunction in male reproductive system cancer patients may be associated with significantly reduced testosterone levels. Testosterone exerts a protective effect against cognitive dysfunction primarily mediated through androgen receptors, involving free radical scavenging and enhanced synaptic plasticity. This is mainly observed in prostate cancer patients undergoing androgen deprivation therapy (ADT) (Holtfrerich et al., 2020; Yan et al., 2019). ADT is a common treatment for patients with localized prostate cancer or those with a rising PSA after definitive therapy without evidence of metastasis. The frontopolar cortex (FPC) is considered crucial for working memory and other cognitive processes, including planning and managing multiple behavioral goals. Chaudhary et al. (2024) found that patients receiving 6 months of ADT exhibited increased cortical thickness (CT) in the FPC. Changes in testosterone levels were correlated with changes in FPC thickness and working memory across all participants.

Multiple studies have found associations between cognitive impairment and anxiety, depression, sleep disturbances, and post-traumatic stress (Boscher et al., 2020; Dhillon et al., 2018; Ng et al., 2018). Furthermore, the immense psychological stress following a cancer diagnosis can lead to declines in both subjective patient evaluations and neuropsychological test performance. Although specific neuroimaging markers for these symptoms are still under exploration, they should be controlled as covariates during analysis to distinguish emotion-related from treatment-specific brain alterations.

Beyond psychological factors, age may also contribute to cognitive decline. Cancer might accelerate normal aging due to increased DNA damage and reduced repair capacity, limiting cognitive reserve and brain reorganization in cancer patients (Mandelblatt et al., 2013). Studies show that compared to untreated patients and controls, older breast cancer patients receiving chemotherapy had poorer baseline cognitive reserve and more severe CRCI, particularly in processing speed (Ahles et al., 2010). Additionally, age-related declines in hormone levels have been found to play a role in CRCI, as anti-hormone therapies can amplify cognitive dysfunction induced by hormonal fluctuations in elderly cancer patients (Harrison et al., 2021). In neuroimaging analysis, age-related brain changes, such as natural hippocampal volume reduction, are strong confounding factors that must be statistically controlled; otherwise, they may mask or exaggerate treatment-specific effects.

Genetics also contribute to a patient’s risk of developing CRCI. Research indicates that genes for catechol-O-methyltransferase (COMT), apolipoprotein E (APOE), and brain-derived neurotrophic factor (BDNF) are closely linked to the occurrence of CRCI. Individuals with these specific genetic variants appear more susceptible to CRCI (Gonzalez et al., 2015; Kautiainen et al., 2023).

Attention is fundamental to cognitive function. Multiple studies report that cancer treatment, particularly chemotherapy, is associated with subjective complaints and objective test performance declines in attention among patients. Neuropsychological studies show impaired attention in breast cancer patients on tasks like the Continuous Performance Test (CPT) post-chemotherapy. Event-related potential (ERP) studies observe reduced amplitude of the P3 (P300) component in these patients (Kam et al., 2016). EEGstudies also find abnormal brain electrical activity in chemotherapy patients (Moore et al., 2014). fMRIresearch indicates altered activation patterns in brain regions like the prefrontal cortex during working memory or attention tasks post-chemotherapy (McDonald et al., 2012). In pediatric ALL patients, Fellah et al. (2019) also found treatment-related alterations in brain activation patterns during attention tasks. Patient self-reported attention problems are common. A large-scale study found a significant increase in subjective cognitive complaints in breast cancer patients after chemotherapy (Janelsins et al., 2017). However, the correlation between subjective cognitive complaints and objective neuropsychological test results is often low, suggesting these methods may reflect different dimensions of cognitive impairment or be influenced by different factors. Kesler (2014) using resting-state fMRI, found altered functional connectivity in brain networks, with the default mode network (DMN) potentially serving as a biomarker in breast cancer patients post-chemotherapy. DTI studies consistently report decreased white matter microstructural integrity in breast cancer patients after chemotherapy, manifested as reduced FA, and this white matter damage correlates with declines in cognitive test performance for attention and processing speed (Deprez et al., 2011; Mzayek et al., 2021). Chen, using arterial spin labeling (ASL), found changes in cerebral blood flow (CBF) in attention network-related brain regions in breast cancer patients receiving neoadjuvant chemotherapy (Chen et al., 2017). Findings suggest that cognitive impairment patterns may differ across cancer types. Research has observed glucose metabolism abnormalities in attention-related brain regions like the thalamus in untreated lung cancer patients (Zhang et al., 2016). In contrast, cognitive impairment in breast cancer patients is often associated with specific chemotherapy regimens, such as those containing anthracyclines (Kesler and Blayney, 2016).

Executive function encompasses cognitive flexibility, working memory, inhibitory control, and planning ability. It is one of the most commonly affected and functionally significant cognitive domains in CRCI. Neuroimaging studies indicate that executive dysfunction is closely associated with structural and functional abnormalities in the prefrontal cortex and its related networks (Lange et al., 2019). Research confirms that cancer treatment correlates with multi-dimensional impairment of executive function. Studies collecting test results like Digit Span and Verbal Fluency in breast cancer patients post-chemotherapy have demonstrated chemotherapy-related executive function impairment, linked to decreased gray matter density in prefrontal regions such as the right middle frontal gyrus (Li et al., 2018). Furthermore, prostate cancer patients receiving ADT show reduced capacity on the n-back task assessing working memory (Chaudhary et al., 2024). These behavioral deficits directly reflect patient-reported declines in quality of life and daily functioning difficulties.

sMRI studies provide crucial anatomical bases for executive function impairment. Voxel-based morphometry (VBM) analysis shows significant reductions in gray matter density in the dorsolateral and ventrolateral prefrontal cortices of breast cancer patients receiving chemotherapy, with this structural change correlating with cumulative chemotherapy dose (Li et al., 2018). fMRI further reveals abnormal neural activity patterns in the prefrontal cortex during working memory tasks in chemotherapy patients. Some studies observe task-related hyperactivation, suggesting a compensatory mechanism reflecting decreased neural processing efficiency, where the brain recruits additional resources to maintain cognitive performance (McDonald et al., 2012).

The execution of executive functions relies on circuitry involving the prefrontal cortex and subcortical structures (e.g., striatum, thalamus) and the coordination of large-scale brain networks. DTI data indicate significantly decreased white matter tract integrity in brain regions like the frontal, parietal, and occipital lobes post-chemotherapy in breast cancer patients, manifested as reduced FA. Moreover, the degree of this white matter damage significantly correlates with declines in performance on cognitive tests of attention, processing speed, and memory (Deprez et al., 2011). Rs-fMRIreveals long-term effects of chemotherapy on brain function at the network level. Studies find significantly weakened functional connectivity within the executive control network (ECN) in breast cancer patients after chemotherapy. This abnormal network connectivity change significantly correlates with poorer performance on neuropsychological tests of executive function (Wang et al., 2016). This finding suggests that functional dysregulation of large-scale brain networks is a key aspect in understanding the neural mechanisms of executive dysfunction. It provides potential neuroimaging biomarkers for future development of preventive cognitive interventions or adjustment of treatment strategies. Clarifying the specific neurotoxic pathways of different treatment modalities is crucial for developing targeted neuroprotective strategies and achieving personalized cognitive risk management.

Episodic memory is particularly impaired in CRCI. Multiple studies show declines in memory indices like delayed recall on neuropsychological tests such as the Auditory Verbal Learning Test (AVLT) in breast cancer patients receiving chemotherapy (Lange et al., 2019; McDonald et al., 2010). sMRI studies have found chemotherapy associated with reduced gray matter volume or density in several brain regions, including the hippocampus. As a core structure for episodic memory formation, structural alterations in the hippocampus are a significant reason for impaired memory encoding, consolidation, or retrieval in CRCI patients (Yao et al., 2023). Animal model research by Winocur et al. (2014) further supports this. Studies show that common chemotherapeutic agents like cyclophosphamide and doxorubicin inhibit hippocampal neurogenesis and correlate with deficits in animals on hippocampus-dependent memory tasks.

Working memory, as a temporary storage system for online information maintenance and manipulation, is also affected in CRCI (Lange et al., 2019). fMRI studies have revealed the neural basis of working memory impairment from the perspective of brain functional activity (McDonald et al., 2010). Research finds altered activation patterns in prefrontal and parietal brain regions associated with working memory when cancer patients perform tasks like the n-back task (Kesler, 2014). Some studies observe that patients need to recruit additional brain resources or exhibit different activation patterns under high cognitive load, which may indicate decreased neural processing efficiency—the brain working harder to maintain performance (Askren et al., 2014). Rs-fMRI studies also find altered functional connectivity within the frontoparietal network responsible for working memory, including connections between the prefrontal and posterior parietal cortices (Wang et al., 2016).

Kesler et al. (2013) using magnetic resonance spectroscopy (MRS), found changed metabolite concentrations in brain regions like the prefrontal cortex in breast cancer patients post-chemotherapy, with elevated choline and myo-inositol and a decreased N-acetylaspartate (NAA)-to-choline rati. These metabolic changes may reflect neuroinflammation, glial activation, or impaired neuronal/axonal integrity and correlate with patients’ subjective memory complaints. PET studies have shown altered glucose metabolism in memory-related brain regions like the posterior cingulate gyrus after ADT (Cherrier et al., 2018). Similarly, DTI shows chemotherapy can impair white matter microstructural integrity. Damage to white matter pathways connecting the hippocampus, such as the fornix and cingulum bundle, may directly disrupt neural circuits for memory information transfer (Follin et al., 2019).

The dynamic evolution of memory impairment presents a complex temporal pattern. Research indicates cognitive deficits may appear during or shortly after chemotherapy. While symptoms improve over time for some patients, cognitive issues persist long-term for others (Janelsins et al., 2017). Recovery potential may be influenced by various factors, including age, baseline cognitive function, and treatment regimen (Lange et al., 2019). Currently, research exploring CRCI management strategies is ongoing. Animal model studies suggest interventions like physical exercise may improve chemotherapy-induced memory deficits by promoting hippocampal neurogenesis and enhancing synaptic plasticity (Winocur et al., 2014). n clinical research, cognitive training and physical exercise have emerged as potential non-pharmacological interventions to help patients cope with cognitive difficulties and improve quality of life (Bray et al., 2017).

Processing speed, a fundamental cognitive process, is commonly impaired in CRCI and likely broadly impacts other cognitive functions. Neuroimaging research provides important insights into this impairment, revealing underlying white matter microstructural alterations and brain network dysfunction (Lange et al., 2019). Patients often show performance declines on standardized neuropsychological tests assessing processing speed, such as Part A of the Trail Making Test and the Digit Symbol Substitution Test (Collins et al., 2014). From a network neuroscience perspective, efficient information processing relies on the overall coordination of brain structural and functional networks. Studies find altered topological properties of brain networks in CRCI patients. Specifically, global efficiency, measuring overall information transfer between different brain regions, tends to decrease. Concurrently, the optimal small-world network property—maintaining tight local connections while enabling efficient long-range communication between different brain regions—is weakened in CRCI patients, indicating their brain networks may deviate from the optimal pattern for information integration and processing (Amidi et al., 2017; Liu et al., 2022). These network-level changes imply less efficient and fluid information exchange between brain regions, explaining the slowing of processing speed.

Compared to memory and processing speed, language dysfunction has received relatively less attention in CRCI research, yet its impact may involve multiple levels from lexical retrieval to pragmatic communication. The Controlled Oral Word Association Test (COWA), included in the core assessment battery recommended by the International Cognition and Cancer Task Force (ICCTF), assesses verbal fluency, indicating lexical retrieval deficits are a measurable dimension of CRCI. Clinical observations and patient self-reports also note problems like word-finding difficulties and comprehension issues in complex situations, potentially affecting daily communication and quality of life (Deprez et al., 2018). Language dysfunction poses a potential threat to patients’ social interaction, occupational ability, and mental health, potentially exacerbating perceived cognitive decline and risk, especially in elderly patients (Lange et al., 2019). Currently, specific research on language impairment in CRCI remains insufficient. Future studies need to utilize multimodal neuroimaging techniques combined with refined language tasks to systematically elucidate its neural mechanisms and provide a basis for developing targeted rehabilitation strategies.

Lim et al. (2016) found reduced glucose metabolism in the medial prefrontal cortex and hippocampus of rats treated with doxorubicin or cyclophosphamide. Barry et al. (2018) found decreased ¹⁸F-FDG uptake in the prefrontal cortex of rats 30 days after doxorubicin treatment, consistent with Lim’s findings. The doxorubicin/cyclophosphamide model simulates the anthracycline/alkylating agent combination chemotherapy commonly used in cancers like breast cancer. Animals receiving a single intraperitoneal injection of doxorubicin or cyclophosphamide showed significant impairment in episodic memory, spatial learning and memory, and contextual fear memory on behavioral tests like novel object recognition, Morris water maze, and fear conditioning. These behavioral phenotypes highly correspond to memory decline and executive dysfunction reported by clinical patients. Behavioral abnormalities occurred 1–4 weeks post-administration and could last months, simulating subacute cognitive impairment in the clinic. Winocur et al.’s (2014) rat antimetabolite model simulates treatments for colorectal cancer, lymphoma, and leukemia. Animals exhibited impairment in spatial working memory, reversal learning, and executive function, resembling the phenotypes of decreased processing speed and cognitive flexibility in clinical patients. Concurrently, mouse methotrexate models are used to simulate cognitive sequelae in pediatric ALL patients after high-dose methotrexate, showing long-term white matter damage and cognitive deficits (Melis et al., 2023). Cisplatin models can induce mitochondrial dysfunction and oxidative stress, with animals showing impaired performance on spatial memory and cognitive flexibility tasks, consistent with clinically observed cisplatin-related cognitive side effects (Lomeli et al., 2017).

Neuroimaging in animal models is primarily used for non-invasive monitoring of dynamic changes in brain structure and function, correlating with behavior. Findings show both overlap and differences with clinical studies.

Animal PET shows significantly reduced glucose metabolism in the medial prefrontal cortex and hippocampus of rats treated with doxorubicin or cyclophosphamide, consistent with clinical ^18^F-FDG PET results. These two brain regions are crucial for human cognition, especially executive function and episodic memory, making their hypometabolism a potentially cross-species consistent imaging marker for CRCI. Additionally, research confirms chemotherapy can inhibit neurogenesis in the hippocampal dentate gyrus (Christie et al., 2012; Dubois et al., 2014). This finding supports indirect clinical imaging observations of hippocampal volume reduction and altered functional connectivit (Feng et al., 2020)^, suggesting the hippocampus is a key target of chemotherapy neurotoxicity, and neurogenesis inhibition is a key mechanism in cognitive impairment. TSPO-PET studies suggest neuroinflammation involvement in CRCI (Schroyen et al., 2022)^, Animal models provide direct histological evidence that chemotherapy activates microglia, increasing pro-inflammatory cytokine expression in the hippocampus and prefrontal cortex, offering an explanation for clinically observed phenomena (Chiu et al., 2017).

However, while clinical DTI studies report widespread decreases in white matter microstructural integrity (Deprez et al., 2011; Menning et al., 2018)^, animal DTI models, although also showing white matter changes, present spatial patterns, severity, and associations with cognitive domains that are difficult to compare directly with human studies. This may be due to fundamental differences in white matter structure complexity, proportion, and function between rodents and humans. Secondly, fMRI studies reveal complex alterations in large-scale brain network connectivity in human (Kesler and Blayney, 2016; Wang et al., 2016). Although functional networks exist in rodents, their lower homology and complexity compared to humans limit direct study of these advanced brain network disorders and their association with specific cognitive symptoms in animal models.

An important feature of CRCI is that cognitive impairment persists for years or even decades in some patients, becoming a chronic sequela. Currently, animal models capable of simulating such long-term progressive impairment are scarce. One study used APOE4 gene-replaced mice treated with doxorubicin to simulate the interaction between genetic risk factors and chemotherapy, observing more persistent cognitive impairment and brain structural changes in aged mice (Demby et al., 2020)^, providing ideas for modeling high-risk subgroups. Overall, current models still have shortcomings in simulating the heterogeneity of human CRCI, its long-term dynamic evolution, and its comorbidities with the chronic disease state of cancer.