|Home | About | Journals | Submit | Contact Us | Français|
We evaluated the course of neurocognitive functioning in newly diagnosed high-grade glioma patients and specifically the effect of tumor recurrence. Following baseline assessment (after surgery and before radiotherapy), neurocognitive functioning was evaluated at 8 and 16 months. Neurocognitive summary measures were calculated to detect possible deficits in the domains of (1) information processing, (2) psychomotor function, (3) attention, (4) verbal memory, (5) working memory, and (6) executive functioning. Repeated-measures analyses of covariance were used to evaluate changes over time. Thirty-six patients were tested at baseline only. Follow-up data were obtained for 32 patients: 14 had a follow-up at 8 months, and 18 had an additional follow-up at 16 months. Between baseline and eight months, patients deteriorated in information-processing capacity, psychomotor speed, and attentional functioning. Further deterioration was observed between 8 and 16 months. Of 32 patients, 15 suffered from tumor recurrence before the eight-month follow-up. Compared with recurrence-free patients, not only did patients with recurrence have lower information-processing capacity, psychomotor speed, and executive functioning, but they also exhibited a more pronounced deterioration between baseline and eight-month follow-up. This difference could be attributed to the use of antiepileptic drugs in the patient group with recurrence. This study showed a marked decline in neurocognitive functioning in HGG patients in the course of their disease. Patients with tumor progression performed worse on neurocognitive tests than did patients without progression, which could be attributed to the use of antiepileptic drugs. The possibility of deleterious effects is important to consider when prescribing antiepileptic drug treatment.
Patients with high-grade glioma (HGG)4 (i.e., primary brain tumors arising from glial tissue) have a poor prognosis. Despite multimodality treatment with surgery, radiotherapy, and chemotherapy, mean survival ranges from around 12 months for patients with glioblastoma multiforme to between one and five years for patients with anaplastic astrocytoma, depending on age and performance status at diagnosis (Hess et al., 2004; Scott et al., 1998; Stupp et al., 2005). Patients with a brain tumor with oligodendroglial characteristics have a somewhat better prognosis (Donahue et al., 1997).
In the course of their disease, most HGG patients are confronted with neurocognitive deficits, subsequently affecting their health-related quality of life (HRQOL) and that of their caregivers. Neurocognitive functioning in glioma patients can be affected by the tumor (Archibald et al., 1994; Hahn et al., 2003; Kaleita et al., 2004; Kayl and Meyers, 2003; Klein et al., 2001, 2003a; Meyers and Hess, 2003; Salander et al., 1995); by tumor-related epilepsy (Klein et al., 2003b; Taphoorn, 2003); by treatment with neurosurgery, radiotherapy, chemotherapy, antiepileptics, or corticosteroids (Crossen et al., 1994; Gregor et al., 1996; Klein et al., 2001, 2002, 2003b; Olson et al., 2000; Sheline et al., 1980; Swennen et al., 2004; Taphoorn and Klein, 2004); and by patient-related factors, including age (Kaleita et al., 2004; Kayl and Meyers, 2003; Klein et al., 2003a, b; Taphoorn and Klein, 2004) and psychological distress (Anderson et al., 1999).
Most studies on neurocognitive function in brain tumor patients pertain to those with low-grade glioma, and only a few studies have collected follow-up data in HGG patients (Archibald et al., 1994; Meyers and Hess, 2003; Taylor et al., 1998). The published studies have generally used a retrospective design, or insensitive screening instruments for this patient population, such as the Mini-Mental State Examination (Taylor et al., 1998).
By using an extensive battery of tests, however, neurocognitive decline could be reliably demonstrated in patients with recurrent HGG even before radiological (MRI) evidence of tumor recurrence (Meyers and Hess, 2003). Considering that neurocognitive status is a predictor of survival in the older population of newly diagnosed HGG patients (Klein et al., 2003a), it is suggested that neurocognitive testing may be a sensitive and informative tool in clinical trials in these patients.
The present study was performed as part of a longitudinal study into neurocognitive function and HRQOL of newly diagnosed HGG patients. The results of these patients’ neurocognitive status at baseline (i.e., following surgery, prior to the start of radiotherapy) have been published (Klein et al., 2001). In that study, neurocognitive function and HRQOL of HGG patients were compared with those of non-small-cell lung cancer (NSCLC) patients and of healthy controls, which thus also accounted for the psychological impact of having cancer. Although HRQOL of HGG patients in the postsurgical period was similar to that of NSCLC patients, glioma patients additionally suffered from a number of condition-specific neurological and neurocognitive problems that significantly affected their daily lives.
In the present study, we present the 8-month and 16-month neurocognitive follow-up of these HGG patients. We excluded the results of the neuropsychological assessments of the NSCLC patients in this study, because the primary objectives here were to evaluate neurocognitive function in HGG patients in the course of their disease and to identify tumor, patient, and treatment characteristics determining neurocognitive functioning.
This study is part of a large, nationwide study in the Netherlands to evaluate the neurocognitive status and HRQOL of HGG patients and low-grade glioma patients and their caregivers. The study was approved by the institutional review boards of all the participating hospitals. The methodology of the study has been described in detail elsewhere (Klein et al., 2001). Briefly, consecutive, newly diagnosed, and histologically confirmed patients who (1) had anaplastic astrocytoma, anaplastic oligodendroglioma, or glioblastoma multiforme, (2) had a life expectancy of at least three months, (3) were eligible for radiotherapy, and (4) were able to communicate in the Dutch language were recruited from the five medical centers listed in the Acknowledgments. Patients were asked to participate by their treating physician (i.e., neurologist, neurosurgeon, medical oncologist, or radiation oncologist), and informed consent was obtained. Medical charts were reviewed to obtain data on patient, tumor, and treatment characteristics.
A battery of standardized tests was used to assess neurocognitive functioning. The tests were administered at baseline, at 8 months, and at 16 months. Completion of the tests required 60 to 90 min. The appendix provides detailed information on the tests used. Neurocognitive performance of glioma patients was compared with the cognitive performance of healthy controls (Jolles et al., 1995), matched with respect to age, gender, and educational level.
With the results of this battery of tests, we calculated neurocognitive summary measures based on the performance of the healthy controls (Z-scores) to detect possible deficits in the glioma patients’ neurocognitive domains of (1) information-processing speed, (2) psychomotor function, (3) attentional functioning, (4) verbal memory, (5) working memory, and (6) executive functioning. Construction of these cognitive domains was based on a principal component analysis using varimax rotation with Kaiser normalization (Kaiser, 1958) performed on the Z-scores (standard equivalents) of the healthy controls. The outcome of the principal component analysis was a confirmation of cognitive domains conventionally used in neuropsychological practice. Individual cognitive test scores of the patients were converted to Z-scores by using the means and standard deviations of the matched healthy controls as a reference. Cognitive domain summary measures (means) based on these Z-scores were calculated for patient groups at baseline, at 8 months, and at 16 months.
According to the number of neurocognitive assessments that could be performed, patients were assigned to one of three groups: (1) patients with a single baseline assessment, (2) patients with a baseline assessment and an 8-month follow-up, and (3) patients with a baseline assessment, an 8-month follow-up, and a 16-month follow-up.
Pearson χ2 tests or analyses of variance (ANOVAs) were used to compare differences between patient groups in sociodemographic and clinical characteristics. Where appropriate, differences in neurocognitive performance between patient groups were also analyzed with analysis of covariance (ANCOVA), correcting for differences in age and educational level. A paired t-test was performed to evaluate the neurocognitive function of HGG patients at baseline and at eight-month follow-up (groups 2 and 3).
To estimate the influence of tumor recurrence on neurocognitive function, the test results of patients with tumor recurrence were evaluated separately with ANCOVAs. Their neurocognitive performance was compared to that of patients without tumor recurrence, correcting for age, educational level, radiotherapy fraction dose, number of fractions, and use of dexamethasone and antiepileptic drugs. Repeated-measures ANCOVAs with group as the between-subject factor were performed to evaluate the differences in neurocognitive performance between glioma patients and those without tumor recurrence over time.
Of the 90 eligible HGG patients, 18 declined to participate because they expected the study to be too stressful, and four patients were excluded because of severe aphasia. Consequently, 68 HGG patients were included. The mean interval between the baseline and the eight-month neurocognitive follow-up was 33 weeks (range, 30–38 weeks), and the mean interval between the 8-month and 16-month follow-up was 31 weeks (range, 23–40 weeks).
Table 1 shows the numbers of HGG patients at each follow-up with neuropsychological assessment. Table 2 lists the patient characteristics. As expected, patients with a 16-month follow-up (group 3) were significantly younger (P = 0.000) and more often had a WHO grade III tumor (P = 0.000) in comparison with patients from groups 1 and 2, who were more often affected by WHO grade IV tumors.
Of the 32 patients with follow-up data, 15 suffered from tumor recurrence during follow-up: 13 patients between baseline and 8-month follow-up assessments and two patients between 8-month and 16-month follow-up assessments (Table 1). Since the number of patients in the latter group was too small to enable statistical analyses, only those patients with tumor recurrence between baseline and eight-month follow-up were evaluated. HGG patients with tumor recurrence during follow-up (n = 13) suffered more often from a grade IV than a grade III glioma (χ2; P = 0.003) and received a lower number of radiotherapeutic fractions (χ2; P = 0.048) with a higher fraction dose (χ2; P = 0.045) when compared with patients without tumor recurrence during follow-up. At baseline, the use of dexamethasone and antiepileptic drugs was equal in both groups, whereas at the eight-month follow-up, HGG patients with tumor recurrence used dexamethasone (χ2; P = 0.005) and antiepileptic drugs (χ2; P = 0.017) more frequently when compared with HGG patients without recurrence (n = 17). One patient with tumor recurrence underwent reoperation, and another was reirradiated. Four of five patients with tumor recurrence were treated with chemotherapy (three patients with a combination of procarbazine, CCNU [1-(2-chloroethyl)-3-cyclohexyl-1-nitrosourea, or lomustine], and vincristine [PCV] and one patient with temozolomide). Two patients without tumor recurrence received chemotherapy (one received PCV in the adjuvant setting, and another received dibromodulcitol and BCNU [N,N-bis(2-chloroethyl)-N-nitrosourea, or carmustine] as part of a trial).
Of the 16 patients with a 16-month follow-up, five had tumor recurrence before the 8-month follow-up (Table 1). The use of dexamethasone and antiepileptic drugs by the two groups of patients did not vary significantly. Four of the five patients with tumor recurrence were treated: Two underwent reoperation, one of whom was also treated with chemotherapy (PCV). One patient was treated only with chemotherapy (PCV), and one underwent reirradiation.
As evidenced by ANCOVAs (correcting for age and educational level), there were no statistically significant differences in neurocognitive status at baseline between the three patient groups with different follow-up lengths. Evaluation of neurocognitive function of patients with and without tumor recurrence during follow-up demonstrated a lower information-processing capacity at baseline in patients with tumor recurrence than in those without recurrence (a mean difference of 1.91 Z-score units [P = 0.036]).
At eight-month follow-up, neurocognitive functioning could be tested in 32 patients, but some did not attend every test of the neuropsychological battery. Figure 1 depicts the neurocognitive performance of this patient group at baseline and at eight-month follow-up. Deterioration in information processing, psychomotor function, and attentional tasks was found. The mean differences for these three domains were 1.19 (P = 0.017), 1.24 (P = 0.041), and 0.89 Z-score units (P = 0.030), respectively.
Regarding individual variations in this group, 18 patients showed a decline in the information-processing tasks (6 patients, >2 Z-score units), whereas 10 showed a slight improvement (9 patients, <1 unit). Concerning psychomotor speed, 14 patients showed a decline within this period (6 patients, >2 Z-score units), and 14 improved (10 patients, <1 Z-score unit; none, >2 units). Regarding attentional tasks, 13 patients deteriorated between baseline and eight-month follow-up (5 patients, >2 units), whereas 12 improved (11 patients, <1 Z-score unit).
The influence of tumor recurrence between baseline and eight-month follow-up on neurocognitive function was estimated by comparing 13 patients with tumor recurrence and 17 without recurrence. Figure 2 depicts neurocognitive performance on each of the six cognitive domains in patients with tumor recurrence and those without tumor recurrence, both at baseline and at eight-month follow-up. At eight months, the patient group with tumor recurrence performed much worse than the “non-progressors” on tests of information processing, psychomotor speed, and executive functioning. The mean differences for these three tests were 4.12 (P = 0.023), 3.42 (P = 0.010), and 2.76 Z-score units (P = 0.018), respectively.
Between baseline and eight-month follow-up, repeated-measures ANCOVAs (correcting for age, educational level, radiotherapeutic fraction dose, and number of fractions) were performed to compare neurocognitive function in patients who had tumor recurrence with that in patients without recurrence. Patients with tumor recurrence showed a decline within this period by a mean of 2.55 Z-score units on information-processing capacity, whereas patients without recurrence showed a mean decline of only 0.35 unit (P = 0.026). On psychomotor function, patients with tumor recurrence showed a mean decline of 2.97 Z-score units, whereas those without recurrence showed a mean decline of 0.09 unit (P = 0.017). On executive-functioning tasks, patients with tumor recurrence deteriorated (mean decline of 1.36 Z-score units), while those without recurrence improved by a mean of 0.61 unit (P = 0.044).
At the eight-month follow-up, patients with tumor recurrence used dexamethasone and antiepileptic drugs more frequently. After correction for this drug use, the neurocognitive function of patients with tumor recurrence did not vary significantly from that of patients without tumor recurrence. Subanalysis showed that the observed differences could be attributed to the use of antiepileptic drugs.
The variations in the individual results for patients with tumor recurrence and for those without tumor recurrence were evaluated. On information-processing tasks, seven patients with tumor recurrence deteriorated (5 patients, >2 Z-score units), whereas 10 patients without recurrence deteriorated (1 patient, >2 units). Four patients with tumor recurrence improved during the first eight months of follow-up (1 patient, >1 Z-score unit). Five patients without tumor recurrence improved (<1 unit) during this period.
On psychomotor-speed tasks, seven patients with tumor recurrence deteriorated (5 patients, >2 units). Six patients without recurrence declined between baseline and eight-month follow-up (1 patient, >2 units). Four patients with tumor recurrence improved within this period (1 patient, >1 unit). Nine patients without tumor recurrence improved during this period (of whom three patients improved by more than 1 unit). None of these patients improved by more than 2 units.
Five patients with tumor recurrence showed a decline on executive-functioning tasks (3 patients, >2 Z-score units). Five without recurrence showed a decline on executive-functioning tasks, but none declined more than 2 units. Five patients with tumor recurrence improved between baseline and eight-month follow-up (1 patient, >1 unit). Eleven without tumor recurrence improved on executive functioning tasks (3 patients, >2 units).
Follow-up data at 16 months were available for 18 patients (group 3). Two patients were excluded from statistical analysis as noted under Patient Characteristics. Of the 16 patients included in this analysis, five had tumor recurrence before the eight-month follow-up (Table 1). Figure 3 depicts the neurocognitive performance on each of the six domains for the patients with and without tumor recurrence (before the 8-month follow-up assessment) at the 8-month and 16-month follow-ups. At 16 months, the patient group with tumor recurrence performed much worse on psychomotor-speed tasks than did the group whose tumor did not progress, with a mean difference of 2.52 units (P = 0.047), although the use of antiepileptic drugs and dexamethasone was equally divided.
In this study, we evaluated neurocognitive functioning in newly diagnosed HGG patients during the course of their disease. Whether, or to what extent, neurocognitive decline occurs during the disease is very important to both patient and caregiver, because these limitations interfere with HRQOL.
Consistent with the findings in the literature, HGG patients with a longer follow-up often had less malignant (grade III) tumors and were younger when compared to patients with a shorter follow-up (Hess et al., 2004; Klein et al., 2003a; Meyers et al., 2000; Scott et al., 1998; Taylor et al., 1998). The neurocognitive performance of the three HGG patient groups did not differ significantly at baseline.
Between baseline and eight-month follow-up, the group as a whole declined in neurocognitive function. This observed deterioration could have been caused by the use of medication, such as antiepileptic drugs (Klein et al., 2003b) or dexamethasone, which were used more often at the eight-month follow-up when compared to baseline, but could also have been caused by other factors such as radiotherapy and tumor progression.
All patients with a follow-up at eight months had been treated with radiotherapy, and the question can be raised of whether this treatment could have interfered with neurocognitive function. Early-delayed radiation encephalopathy may occur between one and six months after completion of radiotherapy. In this period, patients can exhibit more pronounced neurological deficits with accompanying neurocognitive impairment. Although spontaneous recovery with return to normal baseline may occur within 12 months (Sheline et al., 1980), it is feasible to assume that some of our patients suffered from transient radiation-induced encephalopathy at the time of the first follow-up.
Another important factor that may have contributed to neurocognitive decline is the development of tumor recurrence. During follow-up, our HGG patients deteriorated in neurocognitive performance regardless of whether they had tumor recurrence, although this functional decline was much more pronounced in the patient group with tumor recurrence. However, patients with tumor recurrence more often suffered from grade IV tumors and received a lower number of fractions and a higher fraction dose. In our analysis, we corrected for these differences in radiotherapeutic treatment. As expected, the patients with tumor recurrence used dexamethasone and antiepileptic drugs more often, and after correcting for the use of antiepileptic drugs, the decline in the two patient groups did not vary significantly. We therefore postulate that patients with tumor recurrence showed worse neurocognitive functioning, but that the deterioration could probably be attributed to the antiepileptic drugs that these patients received.
Regarding the patients without tumor recurrence, deterioration in neurocognitive functioning was also observed between baseline and eight months. As the use of antiepileptic drugs and corticosteroids in this patient group did not vary during the eight months, antiepileptic drugs and corticosteroids did not contribute to the observed deterioration. However, these patients had no evidence of recurrent tumor growth. Therefore, radiotherapy may have played a role in cognitive deterioration.
At baseline (i.e., before any treatment had started and before radiological evidence of tumor recurrence), the patient group that would develop tumor recurrence during follow-up already performed worse on information-processing tasks than did the patient group without tumor recurrence. As the use of antiepileptic drugs and corticosteroids was equally distributed between these patient groups at baseline, this finding confirms that cognitive deficit is a negative prognostic factor in glioma patients (Curran et al., 1993; Klein et al., 2003a).
Analyses on neurocognitive functioning at 16-month follow-up that were performed for 16 HGG patients showed a further worsening of neurocognitive function.
The present analyses indicate that HGG patients as a group decline in neurocognitive function in the course of the disease. However, evaluation of these functions at the individual level shows that improvement in functioning, albeit small, may also occur. Nevertheless, evaluation of the changes in Z-scores shows that occasional neurocognitive improvements have only little clinical significance (less than 1 Z-score unit). However, the finding that patients whose disease progresses and who deteriorate in neurocognitive functioning do so to a much greater extent (>2 Z-score units) than recurrence-free glioma patients is cause for concern.
In conclusion, HGG patients showed a marked decline in neurocognitive functioning in the course of their disease. Patients with tumor progression performed worse on neurocognitive tests than did patients without tumor progression, but the difference may be mainly attributed to the use of antiepileptic drugs. This result implies careful consideration of the choice of antiepileptic drugs in brain tumor patients.
We thank H. Baaijen, J.S.A. Belderbos, E.B. Bongartz, W. Boogerd, D. González González, M.C.C.M. Hulshof, J.A. Langendijk, I. van der Lee, S. Leenstra, W.F. Luitjes, J.G. Salverda, S.I. Tjahja, and C.A.F. Tulleken for permission to recruit their patients; Astrid Albersen, Wilmy Cleijne, Judith Grit, Nick Guldemond, and Walter Oomen for their invaluable help in tracing and testing patients; and Jaap Lindeboom for his expert advice on neuropsychologic assessment. HGG patients were recruited from the following hospitals: University Medical Center Utrecht (Utrecht) and the Netherlands Cancer Institute/Antoni van Leeuwenhoek Hospital, VU University Medical Center, the Slotervaart Hospital, and the Academic Medical Center of the University of Amsterdam (Amsterdam).
|The Dutch Adult Reading Test (Schmand et al., 1991)||The Dutch version of the New Adult Reading Test provides a measure of premorbid capacity based on verbal ability.|
|Line Bisection Test (Schenkenberg et al., 1980)||A device measuring unilateral neglect, which is usually a sequel of massive right-hemisphere lesions. Noticeable errors are most often made by patients with visual field defects who tend to underestimate the side of the line opposite to the defective field. Outcome measures are horizontal and vertical deviations.|
|Facial Recognition Test (Benton and Van Allen, 1968)||This task was designed to detect impairment in the discrimination of faces, a disorder associated with right-hemisphere lesions. Performance is measured as the number of correct recognitions.|
|Judgment of Line Orientation Test (Woodard et al., 1996)||A test of spatial judgment, also designed to detect right-hemisphere dysfunction, is used in a short form.|
|Letter-Digit Substitution Test (Lezak, 1995)||This test provides a measure of psychomotor performance that is relatively unaffected by intellectual prowess and is suitable for groups with an age range exceeding 60 years. The number of items written down in 90 s is registered, as is the decrease in performance when graphomotor speed is involved.|
|Visual Verbal Learning Test (Lezak, 1995)||This version of the Rey Auditory Verbal Learning Test calls for various aspects of verbal learning and recall. Measures used for analysis are memory performance on trial 1 as indicator of immediate recall, total recall after five trials, delayed recall and recognition after 20 min as indicators of memory consolidation into long-term memory, and a delta score as a measure of learning capacity.|
|Working Memory Task (Sternberg, 1975)||This task is designed to measure the speed of memory processes. The underlying principle is that the extra time needed to complete a test in which there is a stepwise increase in the amount of information to be kept in memory is a measure of the ease at which information is processed in working memory. Capacity is measured by using the slope and intercept as a function of the number of letters kept in working memory.|
|Attention and executive function|
|Stroop Color-Word Test (Lezak, 1995)||This is a selective attention task aimed at measuring interference susceptibility and consists of three subtasks with increasing task complexity.|
|Categoric Word Fluency (Benton, 1968)||This is a simple task requiring the generation of words from semantic categories (e.g., animals) within a limited time.|
|Meander (Lindeboom and Jonker, 1989)||Perservative behavior is quantified by this task derived from the Amsterdam Dementia Screening Test.|
|Concept Shifting Test (Houx and Jolles, 1994)||This test, which has two conditions of complexity, predominately measures functions associated with executive function, especially visual scanning and conceptual tracking. The motor component of this task is measured by three dummy conditions in which no cognitive capacity except for graphomotor speed is required.|
1This study was supported by grant VU96-1155 from the Dutch Cancer Society and by Kapteijnfonds.
4Abbreviations used are as follows: ANCOVA, analysis of covariance; ANOVA, analysis of variance; BCNU, N,N-bis(2-chloroethyl)-N-nitrosourea, or carmustine; CCNU, 1-(2-chloroethyl)-3-cyclohexyl-1-nitrosourea, or lomustine; HGG, high-grade glioma; HRQOL, health-related quality of life; NSCLC, non-small-cell lung cancer; PCV, procarbazine, CCNU, and vincristine.