Journal of Neuro-Oncology

, Volume 108, Issue 2, pp 277–283

Investigation of cognitive impairments in people with brain tumors


    • Laboratory of Cognitive and Behavioral NeurologyFondazione IRCCS Istituto Neurologico “C. Besta”
Baseline Neuropsychological Investigations and Batteries

DOI: 10.1007/s11060-012-0815-6

Cite this article as:
Giovagnoli, A.R. J Neurooncol (2012) 108: 277. doi:10.1007/s11060-012-0815-6


Study of brain tumors (BT) has revealed the importance of cognitive and behavioral assessment to clinical care and prognosis. This paper overviews recent literature, focusing on the main points of interest and current methods, providing recommendations for advancing research. Histological aspects, disease progression, treatment-related neurotoxicity, and co-morbidities determine the cognitive patterns of BT. Mental slowing with prominent executive and memory compromise usually mark the advanced phases of disease, whereas normal cognitive performance or subtle behavioral symptoms characterize the early disease course, irrespective of tumor location. Neurocognitive assessment may indicate brain damage in otherwise neurologically normal patients, explain pathological behavior, and provide reliable measures of outcome, contributing to improving the management of patients. Scarce attention has been devoted to social cognitive deficits which are expected to impair autonomy and relationships. Interest in non-pharmacological treatment of cognitive impairment is a growing area although methodological difficulties persist. Homogeneous patient populations, longitudinal study designs including baseline evaluations, and measurement of the lowest and highest levels of cognitive performance seem indispensable to advancing the study of the cognitive and behavioral changes provoked by BT. Future investigations are also expected to clarify the clinical significance of such changes, their effect on quality of life, and the efficacy of specific rehabilitation treatments.


Cognitive functionBehaviorBrain tumorGliomaRadiotherapySurgeryCognitive rehabilitation


Impairment of cognitive function, for example language, memory, visuospatial perception, time–space orientation, attention, and executive function, occurs in nearly all patients with brain tumors (BT) and eventually compromises their independence [1, 2]. Therefore, assessment of cognitive impairments has become increasingly important to diagnosis and follow-up for clinical and research purposes [3, 4]. Neuropsychological testing is often required to implement the results of neurologic examination and neurophysiological or imaging studies that cannot explain the nature of behavioral alterations. Particular indications, for example the monitoring of multiple treatment, and the psycho-physical limitations of BT patients (e.g., fatigability, depression) place special demands on cognitive assessment. A few questions may guide the choice and planning of measurements: What are the mechanisms of impairment? Which cognitive patterns characterize BT? Who needs detailed evaluation? Which batteries respond to these questions? This paper overviews recent literature, focusing current points of interest, open questions, and directions of research.

Causes and mechanisms of cognitive impairment

Histology, disease progression, treatment-related neurotoxicity, neural reorganization, individual psycho-physical conditions, and co-morbidity, for example epilepsy and cardiovascular failures, contribute to determining the type and severity of cognitive impairment in BT patients [58]. According to recent findings, BT cause not only focal neuron disruption and mass effects but also alterations of brain connectivity [9]. Pathological changes in amplitude and synchronization of low-frequency connectivity, involving different neural networks, have been related to learning and memory deficits [10]. Together with toxic and metabolic insults, such alterations explain the non-focal cognitive patterns of BT, suggesting whole brain dysfunction.

The histology and location of BT are major determinants of onset, progression, and severity of impairment. In low-grade gliomas (LGG), clinical experience indicates mild cognitive deficits at disease onset, usually marked by epileptic seizures [11, 12]. In a pre-surgical investigation, Tucha et al. [13] found impaired executive functions, memory, and attention in 91% of patients, reporting specific tumor-related effects. However, neuropsychological tests are rarely conducted before treatment, preventing definite conclusions about the effect of the tumor. Radiotherapy (RT) may provoke different cognitive changes [14, 15], usually many months after disease onset [6]. According to Klein et al. [16], 1–22 years (a mean of six years) after diagnosis, memory, attention, perception, psychomotor speed, and executive ability were worse for 195 LGG patients than for healthy subjects and for patients with low-grade hematological illnesses. Cognitive decline was worse for irradiated than for non-irradiated patients; RT was associated with cognitive impairment, irrespective of disease duration and total radiation dose, but higher fraction doses (greater than 2 Gy) resulted in more severe deficits. This study also demonstrated that attention and loss of executive function was, in part, related to hemispheric location and antiepileptic drugs (AED) but not to surgery. In 65 out of 195 patients, 6–28 years (a mean of 12 years) after diagnosis, cognitive decline was significantly worse in the irradiated than in the non-irradiated patients, irrespective of fraction dose; the decline of executive function and information processing was related to brain atrophy and white matter changes [6]. By contrast, a study of 20 patients by the North Central Cancer Treatment Group, using a wide range of neuropsychological tests, reported no evidence of cognitive deterioration after RT during three-year follow-up [17]. Another study, in which LGG patients were screened for cognitive performance by the mini mental state examination (MMSE) before and after a mean of 7 years after RT [18], also showed that few patients had cognitive deterioration irrespective of the doses and modalities of radiation. This suggests that the tumor itself is of major importance in determining cognitive deficits, and that RT has additional deleterious effects or leaves the cognitive pattern unchanged [3, 6, 1416]. It is difficult to disentangle the effects of LGG and RT, because tumor growth may outweigh the benefits of RT, and neural reorganization may compensate the effects of BT and RT [19], and concomitant therapy or co-morbidity may induce additional deficits. Worthy of note is that most neuropsychological studies of LGG are retrospective, patient groups and RT schedules are heterogeneous, and baseline evaluations are not performed [3, 5, 2023], preventing precise deductions about RT.

Given increased survival, cognitive assessment has become increasingly important in high-grade glioma (HGG) patients. Steinbach et al. [24] and Hottinger et al. [2] reported the significant prevalence of cognitive deficits in long-term survivors of glioblastoma in Germany and USA. Bosma et al. [3] evaluated, prospectively, the course of cognitive functioning in 32 HGG patients: compared with baseline assessment performed between surgery and RT, the 8 and 16 month follow-ups revealed deterioration in attention, information processing, and psychomotor speed; cognitive decline was more pronounced in patients with tumor recurrence or under medication with corticosteroids and AED. Corn et al. [25] also reported progressive deterioration of cognitive functioning in glioblastoma patients. Brown et al. [26] showed that six, 12, 18, and 24 months after surgery the percentage of long-term survivors with cognitive impairment was stable in the absence of recurrence; a clinically significant decrease in the MMSE scores preceded the radiographic changes of tumor progression. Likewise, Armstrong et al. [27] showed that serial neuropsychological evaluations predicted tumor recurrence. Longitudinal studies suggest that the rate of tumor growth and multiple treatment determine the severity of cognitive impairment, although specific cognitive deficits may relate to tumor location [3, 2628]. Further, in 80 patients with recurrent HGG, Meyers et al. [29] demonstrated that a verbal memory test score was independently and strongly associated with survival, when adjustments were made for age, Karnofsky performance status, and the number and extent of surgical resections.

Metastatic BT treated with whole brain RT or yet untreated may be associated with deficits in motor speed, manual dexterity, memory, and executive functions [30, 31]. In brain lymphoma, attention, executive functions, memory, and motor speed deficits represent prominent clinical manifestations, before and after whole brain RT or methotrexate chemotherapy (CT) [32, 33]. However, after CT, cognitive functions may also significantly improve or remain stable [34].

Surgery may provoke cognitive deficits related to the location and extent of tumor removal [11, 35]. Surgical resection of LGG may cause transient specific deficits [36], with improvement of memory functions supported by the opposite hemisphere [11]; long-term impairments are not usually related to the type of surgery [19]. As for HGG, surgery is rarely a significant cause of cognitive deficits. Actually, reduction of mass effects may be beneficial to cognitive functioning, or prevent further deterioration. In a recent study of patients with high-grade tumors Talacchi et al. [37] observed presurgical deficits in memory and word fluency; although these functions improved after surgery, executive function worsened in relation to tumor size, irrespective of the extent of surgical removal. This suggests that, although BT affect cognitive function, surgery may contribute to improving or maintaining the stability of some abilities.

The cognitive effects of CT, steroids, and AED have not been systematically investigated [38]. The negative effects of CT cannot be easily distinguished from those provoked by RT, although CT usually causes transient symptoms [39]. Higher neurotoxicity has been reported with intra-arterial and intra-thecal CT, cisplatin, or methotrexate, but there is scarce neuropsychological documentation of its severity, typology, and duration [8]. AED mainly affect attention and executive function, although some deficits are counterbalanced by improvement secondary to seizure reduction [11] or they vary independently of the number of AED [40].

Cognitive patterns

The effect of BT location on specific cognitive patterns can vary. In this regard, the disorders caused by BT are usually less severe and localizing than stroke-related impairments [41]. Before surgery, RT, and CT, precise neuroanatomical correlations are inconstant: the functions of the area affected by the tumor may be preserved whereas those supported by the opposite hemisphere may be impaired by mass effects and epileptic discharges [11]. In slow-growing tumors, compensation and substitution neural mechanisms tend to mask focal deficits. In high-grade tumors, focal deficits may be surpassed by confusion, headache, and physical symptoms. After treatment, neuropsychological assessment rarely yields focal deficits. Therefore, clinical-pathological variables, treatment, and emotional-behavioral distress may provoke non-localizing cognitive patterns, manifested as mental slowing, poor psychomotor coordination, “frontal” behavior, personality changes, and memory failures [79, 12].

In the absence of confusion and psychological-behavioral changes, some cognitive patterns may be related to particular brain areas. Frontal tumors are associated with deficits in working memory, the inhibition of interference on ongoing actions, social cognition, risk assessment, decision making, use of external feedback, initiative, abstraction, flexibility, and expression. Temporal tumors may affect naming, verbal fluency, comprehension, memory, semantic competence, and social cognition. Tumors of the diencephalon and corpus callosum may provoke memory failure. Occipital-parietal tumors may impair visuospatial recognition, semantic competence, and social cognition. Tumors of the cerebellum may compromise the capacity to modulate and check the mental operations implicated in a variety of activity (executive function, prosody, grammar, theory of mind, spatial memory) [7, 11, 40, 42, 43]. Patients with malignant BT in the language-dominant hemisphere have more problems with memory, attention, word fluency, and verbal learning than patients with non-dominant hemisphere tumors [4], and have little chance of improving after surgery [44].

Objectives of neuropsychological assessment

Basic objectives, common to different neurological disorders, include identification of individual strengths and deficits and neuroanatomical correlation [37, 40, 45]. For patients with severe disease burden and short life expectancy, time-consuming assessment may uselessly increase fatigue and distress, whereas a few tests may contribute to determining the level of awareness and the reliability of informed consent before investigation or treatment. In single cases or groups, the objectives of neuropsychological testing vary in relation to the phase of disease and treatment. For instance, in the early course of the disease, tailored assessment may characterize the effects of BT in patients with otherwise normal neurology [7]; after diagnosis, the neuropsychological results provide criteria for clinical care and decision making [27, 46] and yield indicators for monitoring postsurgical changes and the effects of treatment that requires specific monitoring.

Some functions, for example autobiographical memory and verbal comprehension, affect patients’ ability to report their quality of life (QOL). Further, cognitive deficits are expected to impair QOL, by reducing the capacity to live flexibly in one’s own environment and to adopt a satisfactory lifestyle [33, 47]. Therefore, neuropsychological testing may contribute to QOL assessment, although a direct association between cognitive deficits and QOL has rarely been investigated [46]. Serial neuropsychological evaluations also support prognosis; test scores may predict survival in patients with HGG [29] or brain metastases [31] and may anticipate tumor recurrence by weeks or months [26, 27, 47]. When assessing QOL and prognosis, study of cognitive abilities may improve the management of patients and the results of clinical trials. Finally, neuropsychological testing may yield information for planning non-pharmacological treatment, for example cognitive rehabilitation or psychological support.

Neuropsychological testing

There is large variability in the prevalence of the cognitive deficits associated with BT, ranging from 29% in LGG to 90% in different tumor types [13, 17, 31]. Differences in neuropsychological measurements, and differences in patients populations, tumor types, and treatments may explain this variability. Table 1 summarizes the criteria commonly used to choose neuropsychological tests and batteries. Multidimensional testing enables comprehensive characterization of the cognitive pattern. One test, even if sensitive to a particular problem and with other functions, cannot yield exhaustive information. Pre-structured tests, for example WAIS and MMSE [17, 18], are inconstantly sensitive to specific changes in patients with BT, whereas tests tailored according to precise clinical or research purposes can provide accurate results [8, 17, 18, 29, 46, 47]. One neuropsychological index cannot reflect multiple cognitive changes and the MMSE is not generally regarded as a comprehensive tool for detection of BT or RT related deficits.
Table 1

Selection criteria for neuropsychological tests and batteries for brain tumor patients

Type and frequency of cognitive impairment

Clinical and research purposes (the complexity and duration of the range of tests must accomplish the specific objectives)

Disease burden

Patient awareness, fatigability, and compliance

Adequate psychometric properties

Ability to detect clinically significant deficits

Congruence with patients reports, everyday activities, and quality of life

Staff burden

Neuropsychological testing should follow standardized procedures, have adequate psychometric properties (content, structure, convergent, and divergent validity; inter-rater and inter-test reliability), and have alternative forms. For BT patients the test measures should also be sensitive to the highest and lowest levels of performance, in different phases of disease and treatment, and be able to detect clinically significant changes (those with practical consequences or effects on everyday activity) that may be different from statistically significant changes (observed at the group level). The neuropsychological results should be contrasted with neurologists’ and patients’ reports of autonomy and QOL, with extrapolation of measurements with practical implications [16, 29, 43, 47].

Given the complexity of the pathogenic mechanisms and the variability of patients’ collaboration, stepwise evaluation, from brief testing to detailed examination, may flexibly satisfy clinical indications, patients’ fatigability, and staff burden. Screening should not surpass 30 min. Test timing, in particular in clinical trials, depends on the study design. In any case, baseline presurgical testing is indispensable to the determination of cognitive changes.

Table 2 summarizes the most sensitive tests used in multidimensional studies [68, 11, 18, 27, 29, 42, 43, 4648].
Table 2

Cognitive tests sensitive to brain tumor and treatment effects [68, 11, 18, 27, 29, 42, 43, 4648]

Trail making test, digit-symbol association, Corsi blocks span (visuomotor coordination speed, set shifting, working memory)

Stroop color-word test, attentive matrices (divided attention, interference control)

Grooved pegboard (motor speed and dexterity)

Raven colored progressive matrices, Wisconsin card sorting test (abstraction, set shifting)

Word fluency on letter and semantic cues, design fluency (initiative, fluency)

Rey complex figure recall, short story recall (episodic memory) California verbal learning test, Rey auditory learning test (learning)

Mini mental state examination

According to Taphoorn and Klein [8], tests for patients with LGG or HGG, adopting a hierarchical model, assessing perception, information processing, attention, executive, memory, and intellectual abilities, would take approximately one hour. By use of a 40-min set of tests assessing cognitive impairment, disability, and QOL, Meyers and Hess [47] showed that tests for working memory, verbal learning, word fluency, attention, and manual dexterity predicted the radiological progression of malignant BT. Inclusion of a few neuropsychological test scores in a regression analysis with disease-related variables also enabled prediction of survival of patients with HGG [29]. Costello et al. [43] showed that specific testing of memory, visuospatial perception, executive, and intellectual abilities distinguished patients with frontal low-grade or non-malignant tumors from patients with frontal high-grade tumors who did not improve after RT. Douw et al. [6] adopted a small group of tests of executive function, mental control, attention, learning, and memory, documenting only attention deficits in long-term survivors of LGG treated by RT. Vigliani et al. [48], using a group of tests for memory, learning, abstract reasoning, attention, and executive function, demonstrated that a reaction time test was able to detect attention deficits 6 months after RT and to monitor long-term tumor progression. In adult patients with high-grade, low-grade, or non-malignant BT, lowest performance was associated with left hemisphere location and tests assessing memory and executive function were most accurate in detecting cognitive deficits (Giovagnoli et al., personal communication). Brown et al. [18] demonstrated that better MMSE scores in HGG patients were associated with longer time to tumor progression and suggested that even generic neuropsychological indices may yield practical information. Fliessbach et al. [49] described computer-based assessment of memory, verbal fluency, attention, and inhibitory control; testing was fast, valid, and reliable, and therefore useful for clinical trials, although executive functions were under-represented. Study of primary central nervous system lymphoma indicated that, as in glioma, attention, executive function, memory, and motor dexterity tests were sensitive measurements of impairment [32, 34]. In lymphoma patients treated by RT, Harder et al. [33] also demonstrated that psychomotor speed, attention, executive functions, memory, and learning were related to white matter changes caused by RT.

According to these studies, tests for attention, executive function, and memory can detect the main BT-related cognitive deficits and, among these, some measures also have clinical and prognostic significance. Recent findings in patients with focal brain damage and epilepsy [40, 50, 51] suggest that frontal and temporal lobe lesions impair the ability to understand others’ mental states, an important prerequisite to social behavior based on complex neural circuits [52]. Patients with tumors in different brain areas are expected to show social cognitive deficits as a consequence of the lesion itself, seizures, or altered brain connectivity [42], suggesting a screening evaluation of such domain.

Non-pharmacological treatment of cognitive deficits

Improved prognosis and attention to QOL have elicited increasing interest in non-pharmacological treatment of the long-term cognitive sequelae of BT. The effects of cognitive rehabilitation using either retraining or compensation strategies have been reported for single cases, small patients groups, or patient–caregiver dyads, with improvement in attention and memory for patients with low or high-grade BT [5355]. Hassler et al. [55] documented much variability in the efficacy of cognitive training based on “holistic” memory empowerment for 11 patients with HGG, although mean group improvement was observed after 12 weeks. Gehring et al. [53] reported improvement of subjectively perceived cognitive function and scores in verbal attention tests for 140 patients with LGG or HGG. There is also some evidence that non-pharmacological treatment is feasible and well-accepted by patients and may have immediate or short-term benefits. However, a lack of control groups, randomized study designs, and neuropsychological baseline or follow-up evaluation prevent precise assessment of the reliability and efficacy of cognitive rehabilitation for BT patients. Considering that cognitive impairment is a major issue for patients with a favorable prognosis, careful neuropsychological measurement of deficits and reserves are substantial prerequisites for non-pharmacological treatment of these patients.


Careful assessment of cognitive function may contribute to clarifying clinical status, providing reliable indicators for clinical care and treatment. According to the literature, recommendations for clinical care may be: to use sensitive not time-consuming tests, to adopt a stepwise evaluation strategy (screening tests followed by detailed evaluation of specific cognitive–behavioral disorders), and to compare the cognitive pattern with neurologic signs and patients’ reports. Cognitive testing should be routinely associated with QOL assessment, with the objective of verifying the reliability of self-reports and extrapolating the impact of cognitive deficits on everyday life. The neuropsychological results, with evaluation of behavior, mood, and QOL, may yield indications for non-pharmacological treatment. Recommendations for advancing research may be: to select homogeneous patient populations according to histology, lesion location, disease duration, and clinical burden, to plan perspective investigations including baseline pre-surgical assessment, to follow adequate test timing, to construct tests sensitive to the lowest and highest levels of performance, and to assess social cognitive ability.

Conflict of interest

The authors declare they have no conflict of interest.

Copyright information

© Springer Science+Business Media, LLC. 2012