Self-reported cancer-related cognitive impairment is associated with perturbed neurotransmission pathways

Patients and settings

This study is part of a larger, longitudinal study of the symptom experience of oncology outpatients receiving chemotherapy (Miaskowski et al. 2014). Eligible patients were ≥ 18 years of age; had a diagnosis of breast, gastrointestinal, gynecological, or lung cancer; had received chemotherapy within the preceding four weeks; were scheduled to receive at least two additional cycles of chemotherapy; were able to read, write, and understand English; and gave written informed consent. Patients were recruited from two Comprehensive Cancer Centers, one Veterans Affairs hospital, and four community-based oncology programs.

Study procedures

The study was approved by the Committee on Human Research at the University of California, San Francisco, and the Institutional Review Board at each of the study sites. Of the 2234 patients approached, 1343 consented to participate (60.1% response rate). The major reason for refusal was being overwhelmed with their cancer treatment. Eligible patients were approached in the infusion unit during their first or second cycle of chemotherapy by a member of the research team to discuss study participation and obtain written informed consent.

Prior to the second or third cycle of chemotherapy, patients completed the AFI (the measure of CRCI in this study) a total of six times over two subsequent cycles of chemotherapy (i.e., prior to chemotherapy administration, approximately one week after chemotherapy administration, approximately two weeks after chemotherapy administration). All of the other measures were done at the enrollment assessment.

At the enrollment assessment, whole blood was collected into PAXgene ribonucleic acid (RNA) stabilization tubes. Total RNA was isolated according to the manufacturer’s standard protocol (Qiagen, USA) using the PaxGene Blood RNA Kit. All of the samples demonstrated a RNA integrity number of ≥ 8 and were retained for gene expression profiling.

Because techniques for quantification of gene expression changed over time, of the 717 patients who provided a blood sample in the parent study, 357 had their samples processed using RNA sequencing (i.e., RNA-seq sample) and 360 had their samples processed using microarray (i.e., microarray sample) technologies (Supplemental Fig. 1). The current study used the gene expression data from patients in the extreme phenotype latent classes that were obtained using the RNA-seq (n = 261) and microarray (n = 247) technologies.

Instruments

Data analysis

Differences in clinical and demographic characteristics

Of the 261 patients in the RNA-seq sample, 48.7% were in the High class and 51.3% were in the Low class (Table 1). Compared to the High class, the Low class was more likely to be female; less likely to be employed; and had a lower annual income. In addition, the Low class had a lower performance status; a higher number of comorbidities; a higher comorbidity burden; a higher MAX2 score; a lower hematocrit; was less likely to have gastrointestinal cancer and more likely to have lung cancer; and was more likely to self-report diagnoses of lung disease, depression, or back pain.

Of the 247 patients the microarray sample, 46.6% were in the High class and 53.4% were in the Low class (Table 2). Compared to the High class, the Low class had fewer years of education; was more likely to be female; less likely to be married or partnered; less likely to be employed; less likely to exercise on a regular basis; and had a lower annual income. In addition, the Low class had a lower performance status; a higher number of comorbidities; a higher comorbidity burden; was more likely to self-report a diagnosis of depression or back pain; and less likely to report no prior cancer treatment.

Demographic and clinical characteristics included in the gene expression analyses

In the logistic regression analysis for the RNA-seq sample (Table 3), ten characteristics were included in the initial model and six were retained in the final model and were used as covariates in the gene expression analysis (i.e., male gender, being employed, MAX2 score, KPS score, SCQ score, self-reported diagnosis of depression). In the logistic regression analysis for the microarray sample, ten characteristics were included in the initial model and seven were retained in the final model and used as covariates in the gene expression analysis (i.e., years of education, being employed, exercise on regular basis, KPS score, self-reported diagnosis of depression, self-reported diagnosis of back pain, type of prior cancer treatment).

Differential gene expression analyses

For the RNA-seq sample, median library threshold size was 9,402,364 reads. Following the application of quality control filters, 13,936 genes were included in the final analysis. The common dispersion was estimated as 0.191, yielding a biological coefficient of variation of 0.438.

For the microarray sample, all of the samples demonstrated good hybridization performance for biotin, background negative, and positive control assays on the arrays. Following quality control filters, 43,933 loci were included in the final analysis.

Pathway impact analyses

For the RNA-seq sample, two surrogate variables were identified and included in the final differential expression model. For the microarray sample, zero surrogate variables were identified. For both samples, a total of 4,529 genes were included in the pathway impact analyses. Across the two samples, 79 KEGG signaling pathways were significantly perturbed at an FDR of < 0.01 (Supplemental Table 1). For this paper, the seven perturbed pathways related to neurotransmission are discussed (Table 4).

Glutamatergic synapse

Glutamine is the amino acid precursor of glutamate and γ-aminobutyric acid (GABA). Given that 90% of neurons have glutamate receptors, it is the most common neurotransmitter within the central nervous system (Baek et al. 2024). In terms of its association with CRCI, in a study of mice treated with a single dose of doxorubicin compared to controls (Thomas et al. 2017), glutamate uptake was 45% slower in the frontal cortex and took 48% longer to clear the dentate gyrus 24 h post administration. The authors concluded that receipt of doxorubicin alters glutamate neurotransmission in nuclei associated with cognitive function within 24 h of chemotherapy administration. In another study that evaluated the acute effects of doxorubicin (Alhowail et al. 2019), a dose-dependent association was found with impairments in long-term potentiation of hippocampal neurons. The authors noted that long-term potentiation in a complex process that is dependent on hippocampal glutamatergic neurotransmission.

GABAergic synapse

GABA is a key modulator of brain function through its regulation of the inhibitory-excitatory balance between ionotropic and metabotropic GABA receptors (McArdle et al. 2023). In terms of CRCI, in a study of mice that used a network-based analysis to identify amino acid signatures in brain regions affected by systemic doxorubicin treatment (Liu et al. 2023), GABA was a potential signature in brain regions associated with cognitive function (e.g., hippocampus, neocortex, prefrontal cortex). In addition, compared to controls, cognitive impairment was observed in the treated mice. Of note, this study identified that doxorubicin treatment altered the levels of other amino acids (i.e., phenylalanine, tyrosine, methionine) in the brain through liver and kidney damage. These authors suggested that the cognitive impairment associated with doxorubicin may be partially due to damage to other organs.

Dopaminergic synapse

Dopamine is a neurotransmitter that is involved in multiple pathways in the brain (e.g., movement, reward) (Juárez Olguín et al. 2016). In a study of rats given 5-fluorouracil (Jarmolowicz et al. 2019), the treatment impaired attentional shifting and caused a decrease in dopamine release. These results suggest that alterations in dopamine release may contribute to CRCI. In a study that evaluated cortical and hippocampal neurochemical changes induced by doxorubicin in rats (Khadrawy et al. 2021), cortical dopamine levels increased significantly after the injection of chemotherapy. The authors concluded that increased dopamine levels may impair cogntive function through multiple pathways (e.g., induction of oxidative stress and/or inflammation).

Serotonergic synapse

The neurotransmitter serotonin regulates a variety of biological functions (e.g., cognition, mood) (Kalinichenko et al. 2024). In a study of rats treated with carboplatin for four weeks (Kaplan et al. 2016), serotonin release was impaired and decrements in spatial learning discrimination were observed. In addition, dopamine release and uptake were impaired. However, overall dopamine content and the reserve pool of dopamine did not change significantly. Taken together, these results suggest that chemotherapy selectively impairs dopamine release and uptake processes and serotonin release that contribute to impairments in learning.

Additional pathways

While in this study, both the long-term depression and cholinergic pathways were perturbed, no studies were identified that reported on associations between CRCI and either of these pathways. However, in a study that aimed to comprehensively understand the context and distribution of pathways that contribute to Alzheimer’s disease (Morgan et al. 2022), text-mining was used to generate a systematic assessment of the breadth and diversity of biological pathways within a corpus of 206,324 dementia-related publications. Across research spanning 30 years, the long-term depression and cholinergic synapse pathways were consistently associated with Alzheimer’s disease, which suggests that these two pathways are important regulators of brain function.

The endocannabinoid system regulates retrograde signaling in the central nervous system through the modulation of neurotransmitters and synaptic plasticity (Kibret et al. 2023). No studies were identified that reported on associations between CRCI and the retrograde endocannabinoid signaling pathway. However, in a study that evaluated for associations between differential serum metabolites and their corresponding pathways in patients with Alzheimer’s disease (Yu et al. 2023), this pathway was enhanced and the GABAergic synapse pathway was downregulated.

Associations between CRCI and neurotransmission-related genes

While an evaluation of individual genes within these perturbed pathways is beyond the scope of the current study, it is worth noting that evidence from other studies suggests that a variety of neurotransmitter-related genes may contribute to CRCI. For example, catechol-O-methyltransferase (COMT) is an enzyme responsible for regulation of catecholamine neurotransmitters (e.g., dopamine) (Qayyum et al. 2015). In a study of patients with breast cancer (Cheng et al. 2016), compared to patients with the GG genotype for COMT rs165599, patients with the GA and AA genotypes had a lower odds of developing self-reported cognitive decline. However, in two studies of patients with cancer (Lengacher et al. 2015; Barratt et al. 2015), no associations were found between subjective or objective measures of CRCI and COMT rs4860.

The solute carrier 6 family group of genes includes transporters for a variety of neurotransmitters (e.g., GABA, serotonin, dopamine) (Bröer and Gether 2012). In a study of breast cancer survivors (Lengacher et al. 2015), individuals who were homozygous for the common G allele (GG) in solute carrier family 6 member 4 (SLC6A) rs16965628 had greater improvements in three cognitive outcomes (i.e., memory, organization, global cognition) compared to individuals who were heterozygous or homozygous for the rare C allele (GC or CC).

Finally, in a study of patients with breast or prostate cancer (Harris et al. 2023), associations were found between the severity of three distinct symptom clusters (i.e., sickness-behavior, mood-cognitive, treatment-related) and polymorphisms for 16 genes involved in neurotransmission. While CRCI was not evaluated as an individual symptom, associations were reported between the mood-cognitive cluster and genes involved in catecholaminergic (i.e., adrenoreceptor alpha 1D), GABAergic (SLC6A1), and serotonergic (SLC6A2, SLC6A3, 5-hydroxytryptamine receptor (HTR) 2 A, and HTR3A) neurotransmission. Overall, these studies provide additional support for our data-driven hypothesis that alterations in neurotransmission contribute to CRCI.

Limitations

Because patients in this study had a diagnosis of breast, gastrointestinal, gynecological, or lung cancer, our findings may not generalize to patients with other types of cancer. In addition, because an evaluation cohort was not used, these findings warrant confirmation in an independent sample. Because CRCI was assessed using a self-report measure, an evaluation using objective measures may yield different CRCI profiles. Although our CRCI phenotype was created based on longitudinal assessments, blood was collected only at the enrollment assessment. Longitudinal studies are needed that collect both symptom and molecular data to determine if pathway perturbations change over time and to determine the direction for these associations. Equally important, future studies should evaluate for pathway perturbations in other tissues (e.g., cerebral spinal fluid) and/or for associations with additional biomarkers (e.g., genetic, epigenetic).

Competing interests

The authors declare that they have no conflicts of interest.

Ethical approval

This study adhered to the principles of the Declaration of Helsinki. The study was approved by the Institutional Review Board at the University of California, San Francisco and at each of the study sites.

Consent

All of the patients signed written informed consent.