Abstract
The impact of chemotherapy on brain functionality has been widely investigated from a clinical perspective, and there is a consensus on a significant impairment of multiple cognitive domains affecting cancer patients after treatment. Nuclear medicine offers a variety of biomarkers for evaluating possible effects of chemotherapy on the brain and for depicting brain changes after chemotherapy. This review summarizes the most relevant findings on brain imaging in patients undergoing chemotherapy for the most common oncologic diseases. The literature published to date offers exciting results on several radiolabeled compounds, from the more common imaging of glucose metabolism to neuroinflammation. This review also provides a general overview of the literature concerning clinical features and the physiopathologic basis of chemotherapy-related cognitive impairment.
Chemotherapy is associated with debilitating side effects that affect quality of life (1,2). The term chemotherapy-related cognitive impairment (CRCI) describes a clinical condition characterized by memory and concentration impairment, difficulties in information processing and executive functions, and mood and anxiety disorders (3,4), with a highly variable prevalence estimated to range from 17% to 75% (5). There is evidence that chemotherapy drugs such as cisplatin, carboplatin, paclitaxel, cyclophosphamide, vincristine, and lenalidomide are neurotoxic (6): at a molecular level, cytokine dysregulation and oxygen radical production are suspected to be responsible for CRCI (7). Despite several studies on CRCI, there is no consensus on whether specific brain areas are implicated (8). In recent years, molecular neuroimaging techniques have revealed interesting aspects of the underlying mechanisms of CRCI. This review highlights the contribution of neuroimaging to this field, underlining findings and information from the most important studies.
CLINICAL FEATURES OF CRCI
Recognition of an association between cancer-related treatment and cognitive changes in long-term survivors is not something new, with the first reports on this topic dating to the 1980s and 1990s (9). CRCI may have a wide spectrum of symptoms, ranging from problems with attention, concentration, and working memory to problems with executive function (Table 1) (10).
Cognitive Domains and Abilities More Commonly Involved in CRCI
According to longitudinal neuropsychological studies, up to 35% of patients are affected by cognitive impairment months or years after completion of oncologic therapy (11). However, CRCI can affect distinct populations of cancer patients differently according to the differences in tumor histology, biologic behavior, location, and growth rate (12).
There is no consensus on the preferred tools for diagnosing and measuring cognitive impairment in cancer patients who have undergone chemotherapy. However, according to the existing literature, the 2 main methods of assessment in addition to neuroimaging are neuropsychological testing and self-reports of cognitive impairment. Regarding the former method, a wide range of neuropsychological batteries has been recommended (i.e., the Hopkins Verbal Learning Test, Revised; the Trail Making Test; the testing recommendations of the International Cognition and Cancer Task Force; and the Controlled Oral Word Association [part of the Multilingual Aphasia Examination]). The latter method is based mainly on a patient’s subjective perception as measured through tools such as the European Organization for Research and Treatment of Cancer core quality-of-life questionnaire (QLQ-C30) or the Functional Assessment of Cancer Therapy–Cognitive Function Questionnaire.
CRCI PHYSIOPATHOLOGIC BASIS
Cognitive impairment may affect up to 50% of patients undergoing chemotherapy (13). Chemotherapy may lead to encephalopathy with highly complex and heterogeneous molecular mechanisms. The damage induced by chemotherapy affects neurons and microglia. Drug-induced damage to neurons elicits a cascade of events culminating in activation of microglia and astrocytes and disruption of the normal homeostatic relationship between myelinating cells (oligodendrocytes) and oligodendrocyte precursor cells (13). In addition, there is an increase of proinflammatory cytokines that, together with activation of astrocytes, promotes the release of paracrine factors, significantly hampering the maturation of oligodendrocytes (13). Direct involvement of chemotherapy in the release of cytokines needs to be further investigated. Instead of a local increase, it is more probable that circulating cytokines induced by chemotherapy penetrate the blood–brain barrier to directly act on the central nervous system, activating microglia and astrocytes to secrete further cytokines (14). Chemotherapy may affect brain tissue by modifying the shape of neurons, neurotransmitter release, and blood–brain barrier integrity (1) and may slow neurotransmitter uptake and release into neurons (15–17). These finding may partially explain some of the clinical features of CRCI, particularly those related to alteration in emotion, learning, and memory.
SEARCH STRATEGY
Two separate literature searches of the PubMed/Medline databases were performed according to the PRISMA guidelines (Preferred Reporting Items for Systematic Reviews and Metaanalyses). The first assessed the effects of the most commonly used chemotherapeutic drugs on cognitive function, and the second assessed the results of available SPECT and PET examinations of the field of interest.
For search 1, the terms used were a combination of the most commonly used chemotherapeutic agents (i.e., cisplatin, carboplatin, oxaliplatin, cyclophosphamide, methotrexate, fluorouracil, doxorubicin, etoposide, irinotecan, taxanes) AND “chemotherapy-related cognitive impairment” OR “chemobrain.” The following types of studies were considered: head-to-head comparative series, matched-pair studies, clinical trials, case series, prospective studies, and retrospective cohorts. Case reports, conference proceedings, editorial commentaries, interesting images, and letters to the editor were excluded. We selected only studies published from 2012 to June 2022, limited to humans and in the English language, with a cohort of at least 20 enrolled patients (Supplemental Fig. 1; supplemental materials are available at http://jnm.snmjournals.org).
NOTEWORTHY
Brain 18F-FDG PET imaging shows a reduction of glucose metabolism in CRCI.
Chemotherapy may affect DAT receptor expression in the brain.
Imaging of TSPO is a promising tool for the investigation of CRCI.
For search 2, we used a combination of the following terms: “PET” OR “PET/CT” OR “single photon emission tomography” OR “SPECT” OR “translocator protein” OR “dopamine transporter imaging” AND “chemotherapy-related cognitive impairment.” Studies were selected in the same way as for search 1; nevertheless, because of a shortage of published studies on this topic, we decided not to apply any temporal filter and to include studies with at least 10 patients (Supplemental Fig. 2).
Two reviewers conducted the literature search and independently appraised each study using a standard protocol and data extraction. The reference lists of the selected studies were carefully checked to identify any additional relevant literature.
For search 1 (chemotherapeutic agents and CRCI), the extracted data were type of study (e.g., prospective or retrospective), year and location of study, sample size, tumor, type of chemotherapy, and timing of CRCI assessment with regard to chemotherapy completion. For search 2 (PET and SPECT imaging), the extracted data were type (e.g., prospective or retrospective), year and location of the study, sample size, radiopharmaceuticals, device (SPECT or PET only or hybrid devices), modality of image assessment (qualitative or quantitative), type of chemotherapeutic agent, and the eventually performed neuropsychological tests.
Studies with incomplete technical or clinical data were considered ineligible. If studies were by the same group of researchers, only the study with the highest number of enrolled patients was considered. We resolved any discrepancy by discussion. As this was not a metaanalysis, no statistical analysis was performed.
RESULTS
In total, 142 nonduplicate studies were retrieved from the database for search 1 and 30 for search 2. After removal of duplicate records and screening based on title and abstract, the remaining studies underwent full-text eligibility assessment, which identified 22 relevant studies (search 1, n = 14; search 2, n = 8). The main findings of the selected studies for searches 1 and 2 are summarized in Supplemental Table 1 and Table 2, respectively.
Studies on PET and SPECT in CRCI
Chemotherapeutic Agents and CRCI
Fourteen studies, encompassing 2,390 patients, on the effects of various chemotherapeutic agents on cognitive function were selected (18–31). Great heterogeneity in study design was registered, since only 4 of the 14 studies (28.5%) were prospective and 1 (7.1%) was a randomized trial, whereas the remaining 9 (71.4%) were retrospective observational studies. Great variability in timing and modality of CRCI assessment was also noted. Additionally, the time point of CRCI assessment meaningfully varied among selected studies, ranging from interim assessment during chemotherapy cycles to 20 y after therapy completion.
The majority of the selected studies were on breast cancer (53.3%), most probably because of the relevant advances in prevention, diagnosis, and therapy in this field and the relatively good prognosis of this type of cancer in comparison to other types, with 80% of women with primary breast cancer surviving for at least 10 y after mastectomy or breast-conserving surgery (32).
In 5 cases (35.7%), taxanes were used alone or in combination with other chemotherapeutic agents, whereas in 4 cases (28.5%), patients underwent platin-based therapies. Except for 1 study that analyzed the potential impact of treatment with rituximab, cyclophosphamide, hydroxydaunomycin, vincristine sulfate, and prednisone on cognitive function (21), none of the analyzed studies was on recently implemented immunotherapeutic regimens, either those in which monoclonal antibodies targeted tumor-associated biomarkers or those aimed at removing negative immune regulation (33).
All but 1 selected study reported a decline in cognitive function after chemotherapy: in particular, 4 studies (28.5%) (20,23,26,27,29,31,34) reported a more relevant impairment of executive functions, 3 studies (21.4%) (27) documented self-reported or self-perceived cognitive impairment or mood changes (anxiety, trouble sleeping), and 1 study (7.1%) (24) reported reduced social attainment and poor quality of life in cancer survivors (Supplemental Fig. 3).
PET and SPECT Imaging of CRCI
We selected 8 studies concerning the application of SPECT or PET in CRCI and including 198 patients overall. Only 3 tumor types were evaluated: lymphoma (4 studies, 50%), breast cancer (3 studies, 37.5%), and acute myeloid/lymphoid leukemia (1 study, 12.5%); SPECT was used in 2 studies (24,35), whereas PET/CT was used in the remaining 6 studies. As expected because of its availability and its capacity to give accurate insight into brain metabolism, 18F-FDG was the most commonly used radiopharmaceutical (6 studies, 75%). A meaningful heterogeneity in PET evaluation was noted since authors used, aside from qualitative evaluation, quantitative analysis by volume of interest in 4 studies (36–39) and a statistical parametric map in 2 studies (40,41). As regards SPECT studies, one used visual image evaluation (42) and another used quantitative analysis of dopamine transporter (DAT) density (35).
Finally, only a minority (3 studies, 37.5%) (35,39,42) reported a correlation between imaging findings and neuropsychological assessment. Selected studies focusing on PET or SPECT imaging are discussed here.
SPECT Tracers for CRCI Imaging
Cerebral Blood Flow Measurement
Chemotherapy-induced microvascular damage in the brain is thought to be a causative factor of CRCI (43). SPECT with 99mTc-hexamethyl-propylene-amine oxime has been used to assess changes in cerebral blood flow (44), despite its well-known limitations in terms of spatial resolution and quantitation accuracy.
In a cohort of 12 pediatric patients, treated with high-dose cytarabine for acute myeloid leukemia (n = 11) or acute lymphoid leukemia (n = 1), Véra et al. investigated changes in cerebral blood flow through 99mTc-hexamethyl-propylene-amine oxime SPECT after the induction phase, immediately after the first intensification, and during follow-up (42). SPECT results were correlated with neuropsychological assessment and serial brain MRI. At the induction phase, brain SPECT was performed in 8 patients, with slightly heterogeneous findings in 4 and normal findings in the remaining 4, whereas in all 8 examined patients brain MRI findings were normal. At the high-dose consolidation phase, 5 patients had chemotherapy-related neurotoxicity: in such cases, MRI findings were normal in 4 of 5 patients, whereas SPECT findings were diffusely heterogeneous in 4 of 5 patients and slightly heterogeneous in 1 patient. Follow-up was available for 4 patients with neurotoxicity (n = 5); all regressed over time. One patient had particularly prolonged (60 mo) neurologic symptoms associated with persistent abnormalities on SPECT and brain MRI.
DAT Imaging
Molecular imaging of DATs through SPECT has been extensively applied in clinical practice to diagnose and monitor Parkinson disease and other extrapyramidal syndromes (45,46). SPECT with 99mTc-TRODAT-1 (2-[2-[[(1R,2R,3S,5S)-3-(4-chlorophenyl)-8-methyl-8-azabicyclo[3.2.1]octan-2-yl]methyl-(2-sulfidoethyl)amino]ethylazanidyl]ethanethiolate;oxygen(2-);technetium(5+)) was performed by Vitor et al. to investigate DAT integrity in 28 women reporting cognitive impairment related to chemotherapy for breast cancer and in 22 healthy female controls matched for age and level of instruction (35). Calculation of tracer concentration in the striatum by dedicated software (DaTQUANT; GE Healthcare) showed significantly less striatal uptake (both at the analysis of the overall striatum and at the separate analysis of the caudate and putamen) in breast cancer patients experiencing CRCI than in healthy controls, indicating that toxic damage to the basal ganglia may be involved in the complex mechanisms leading to CRCI. Patients developing parkinsonism after chemotherapy are generally characterized by strong responsiveness to levodopa and tend to improve over time (Supplemental Fig. 4) (47).
18F-FDG PET Imaging in CRCI
When approaching PET imaging of the brain using 18F-FDG, one should consider that under normal conditions brain uses glucose as its sole source of energy (48), that hypometabolism may not correspond to areas with the greatest changes in routine neuropathology (49), that hypometabolism is not directly affected by intracellular or extracellular inclusions (50), and that glucose metabolism primarily reflects synaptic activity (51).
Few studies have investigated the potential role of 18F-FDG in patients with CRCI. One of the first was in 2015, on 49 patients with Hodgkin disease (40), who were evaluated at diagnosis and during treatment. Surprisingly, the authors reported a significantly higher metabolic activity after the first cycles in the right angular gyrus (Brodmann area 39) whereas a significant metabolic reduction was found in Brodmann areas 10, 11, and 32 bilaterally. All these changes disappeared at the end of the therapy course (40). The authors concluded that the results are consistent with a transient and limited impact of chemotherapy on brain metabolism in Hodgkin lymphoma. In agreement with the previous report (40), Shrot et al. found increased 18F-FDG uptake in the parietal and cingulate cortices in 14 pediatric patients diagnosed with lymphoma; decreased 18F-FDG uptake was found in deep gray matter nuclei and the brain stem (52). Tauty et al. and Goldfarb et al. found reduced metabolism bilaterally in the anterior cingulate cortex and left inferior frontal and insular cortex soon after 2 cycles of chemotherapy and hypometabolic areas in the left anterior cingulate cortex, in the left inferior frontal and insular cortex, and finally in the left temporal lobe after 6 cycles (53,54). A reduction of glucose metabolism was found in the frontal, cingulate, and temporoinsular regions after 2 cycles of chemotherapy (53). The differences from other studies cited previously may be partially explained by differences in the chemotherapy agents used. In non-Hodgkin lymphoma, a general reduction of around 20% in overall cerebral cortical metabolism was found after chemotherapy (36). These findings suggest a diffuse and severe impairment of brain functionality after chemotherapy in these patients (36). In a population of 10 patients with breast cancer, Ponto et al. found reduced metabolism bilaterally in orbital frontal regions as compared with healthy subjects (37). Interestingly, this finding is consistent with cognition and executive function impairment found by neuropsychological assessment (37).
PET with Tracers Other Than 18F-FDG: Translocator Protein (TSPO) Ligands for CRCI Imaging
In recent years, TSPO, an 18-kDa protein expressed mainly on the outer mitochondrial membrane of several cells of the central nervous system (microglia, astrocytes, and endothelial cells), has emerged as a target for molecular imaging of neuroinflammation, since its expression is minimal in the healthy brain but strong when microglia are activated in response to injury (55). The development of positron-emitting ligands selectively binding to TSPO has allowed in vivo assessment of neuroinflammation through PET/CT or PET/MRI technology (56,57). Notably, binding of second-generation tracers to TSPO is strongly influenced by a single polymorphism (rs6971) in exon 4 of the TSPO gene (58), according to which patients can be stratified into 3 categories: high-, mixed-, and low-affinity binders.
Schroyen et al. have recently investigated the potential of TSPO PET for in vivo imaging of neuroinflammation in breast cancer patients undergoing chemotherapy, also through correlation of PET findings, neuropsychological tests, and inflammatory markers (59). The authors prospectively enrolled patients distributed into 3 different cohorts: breast cancer patients undergoing chemotherapy, breast cancer patients not scheduled for chemotherapy, and a control group of healthy women. The chemotherapy cohort exhibited higher 18F-DPA714 (N,N-diethyl-2-[4-(2-fluoroethoxy)phenyl]-5,7-dimethylpyrazolo[1,5-a]pyrimidine-3-acetamide) uptake in the occipital and parietal lobes than did chemotherapy-naïve and healthy controls. Furthermore, patients undergoing chemotherapy showed altered neuropsychological test scores and increased inflammatory markers compared with the other 2 cohorts. Among inflammatory biomarkers, neurofilament light-chain protein, an axonal damage indicator, was particularly increased in chemotherapy patients and strongly correlated with TSPO PET findings. Despite 18F-DPA714 incorporation, patients undergoing chemotherapy did not show relevant alterations of micro- or macrostructure in white matter on MRI, as determined through pixel-based analysis of diffusion-weighted images.
CONCLUSIONS AND FUTURE OUTLOOK
Nuclear medicine techniques are not commonly considered in the work-up of patients with CRCI-related manifestations, despite their high potential to investigate different physiopathologic phenomena (i.e., cortical metabolism, DAT integrity, and neuroinflammation) through specific imaging probes (60). From careful analysis of the selected studies, some observations can be made on the role of functional neuroimaging in CRCI.
First, few studies have explored the usefulness of 18F-FDG PET/CT for imaging of CRCI, and even fewer studies have used statistical parametric mapping to assess changes in cortical metabolism before and after chemotherapy. We therefore suggest to implement statistical parametric mapping in future clinical trials on the use of 18F-FDG for CRCI imaging. This parametric analysis entails voxel-level statistical parametric mapping at the whole-brain level, comparing each patient with a reference group using a 2-sample t test, thus generating a contrast t-map for areas of relative hypometabolism in the study group compared with the controls (61). In this respect, technologic innovations may be of great value to further improve the imaging approach to CRCI: hybrid PET/MRI, as an example, is a still-underexplored tool for correlating eventual changes in T1- or T2-weighted images with metabolic abnormalities detected by 18F-FDG. In addition, the emergence of artificial intelligence and radiomics may find interesting applications to extract potentially useful data, undetectable by visual evaluation, from 18F-FDG PET image texture analysis (62). PET tracers other than 18F-FDG, such as TSPO ligands, can provide an interesting opportunity to investigate in vivo, at a molecular level, the inflammatory landscape associated with CRCI, but their widespread use is still hampered by high cost, a lack of authorized compounds, and the dependence of image quality on genetic polymorphism.
Another issue is when and how to assess CRCI after therapy completion. An objective determination of chemotherapy effects on cognitive abilities is hampered by the multifaceted nature of CRCI, since underlying depression or anxiety disorders can also be responsible for some symptoms and are often classified as CRCI. The concept of cancer-related posttraumatic stress disorder, a complex set of symptoms affecting patients’ psychosocial and physical well-being during cancer treatment and into survivorship (63), has been gaining ground and should be considered in future clinical trials aimed at applying functional imaging to CRCI investigation.
The impact of chemotherapy on brain has been assessed by a few imaging studies. Moreover, the methodology used in the studies, the most relevant of which are cited in this review, is characterized by a huge heterogeneity in imaging modalities, clinical evaluations, chemotherapies, and types of tumor. Moreover, there is a lack of longitudinal studies to investigate the possible reversible effect of neuronal impairment induced by chemotherapy.
Nuclear medicine offers several instruments for the detailed evaluation of physiopathologic processes underlying CRCI. The research performed for this review indicates that the major constraint on discoveries will be due not to the available techniques of functional imaging, which are constantly improving, but rather to the precision of our hypotheses and the creativity of our methods for testing them. In particular, longitudinal and standardized studies are needed to investigate the impact of each chemotherapy drug on the brain (considering, in particular, the same dose) or combinations of these drugs. Moreover, correlation with a standardized neuropsychological assessment is mandatory to exclude the possible contribution of stress and emotions, especially for functional studies.
DISCLOSURE
No potential conflict of interest relevant to this article was reported.
Footnotes
Published online Feb. 2, 2023.
- © 2023 by the Society of Nuclear Medicine and Molecular Imaging.
REFERENCES
- Received for publication May 14, 2022.
- Revision received January 26, 2023.
In this issue
Jump to section
Related Articles
Cited By...
- No citing articles found.